Tech Talk

Since the first mobile application (app) was developed in the 1990s, apps have become increasingly integrated into medical practice and training. More than 5.5 million apps were downloadable in 2019,¹ of which more than 300,000 were health related.² In the United States, more than 80% of physicians reported using smartphones for professional purposes in 2016.³ As the complexity of apps and their purpose of use has evolved, regulatory bodies have not adapted adequately to monitor apps that have broad-reaching consequences in medicine.

We review the primary literature on PubMed behind health-related apps that impact dermatologists as well as the government regulation of these apps, with a focus on the 3 most prevalent dermatology-related apps used by dermatology residents in the United States: VisualDx, UpToDate, and Mohs Surgery Appropriate Use Criteria. This prevalence is according to a survey emailed to all dermatology residents in the United States by the American Academy of Dermatology (AAD) in 2019 (unpublished data).

VisualDx

VisualDx, which aims to improve diagnostic accuracy and patient safety, contains peer-reviewed data and more than 32,000 images of dermatologic conditions. The editorial board includes more than 50 physicians. It provides opportunities for continuing medical education credit, is used in more than 2300 medical settings, and costs $399.99 annually for a subscription with partial features. Prior to the launch of the app in 2010, some health science professionals noted that the website version lacked references to primary sources.⁴ The same issue carried over to the app, which has evolved to offer artificial intelligence (AI) analysis of photographed skin lesions. However, there are no peer-reviewed publications showing positive impact of the app on diagnostic skills among dermatology residents or on patient outcomes.

UpToDate

UpToDate is a web-based database created in the early 1990s. A corresponding app was created around 2010. Both internal and independent research has demonstrated improved outcomes, and the app is advertised as the only clinical decision support resource associated with improved outcomes, as shown in more than 80 publications.⁵ UpToDate covers more than 11,800 medical topics and contains more than 35,000 graphics. It cites primary sources and uses a published system for grading recommendation strength and evidence quality. The data are processed and produced by a team of more than 7100 physicians as authors, editors, and reviewers. The platform grants continuing medical education credit and is used by more than 1.9 million clinicians in more than 190 countries. A 1-year subscription for an individual US-based physician costs $559. An observational study assessed UpToDate articles for potential conflicts of interest between authors and their recommendations. Of the 6 articles that met inclusion criteria of discussing management of medical conditions that have controversial or mostly brand-name treatment options, all had conflicts of interest, such as naming drugs from companies with which the authors and/or editors had financial relationships.⁶

Mohs Surgery Appropriate Use Criteria

The Mohs Surgery Appropriate Use Criteria app is a free clinical decision-making tool based on a consensus statement published in 2012 by the AAD, American College of Mohs Surgery, American Society for Dermatologic Surgery Association, and American Society for Mohs Surgery.⁷ It helps guide management of more than 200 dermatologic scenarios. Critique has been made that the criteria are partly based on expert opinion and data largely from the United States and has not been revised to incorporate newer data.⁸ There are no publications regarding the app itself.

Regulation of Health-Related Apps

Health-related apps that are designed for utilization by health care providers can be a valuable tool. However, given their prevalence, cost, and potential impact on patient lives, these apps should be well regulated and researched. The general paucity of peer-reviewed literature demonstrating the utility, safety, quality, and accuracy of health-related apps commonly used by providers is a reflection of insufficient mobile health regulation in the United States.

There are 3 primary government agencies responsible for regulating mobile medical apps: the US Food and Drug Administration (FDA), Federal Trade Commission, and Office for Civil Rights.⁹ The FDA does not regulate all medical devices. Apps intended for use in the diagnosis, cure, mitigation, prevention, or treatment of a disease or condition are considered to be medical devices.¹⁰ The FDA regulates those apps only if they are judged to pose more than minimal risk. Apps that are designed only to provide easy access to information related to health conditions or treatment are considered to be minimal risk but can develop into a different risk level such as by offering AI.¹¹ Although the FDA does update its approach to medical devices, including apps and AI- and machine learning–based software, the rate and direction of update has not kept pace with the rapid evolution of apps.¹² In 2019, the FDA began piloting a precertification program that grants long-term approval to organizations that develop apps instead of reviewing each app product individually.¹³ This decrease in premarket oversight is intended to expedite innovation with the hopeful upside of improving patient outcomes but is inconsistent, with the FDA still reviewing other types of medical devices individually.

For apps that are already in use, the Federal Trade Commission only gets involved in response to deceptive or unfair acts or practices relating to privacy, data security, and false or misleading claims about safety or performance. It may be more beneficial for consumers if those apps had a more stringent initial approval process. The Office for Civil Rights enforces the Health Insurance Portability and Accountability Act when relevant to apps.

Nongovernment agencies also are involved in app regulation. The FDA believes sharing more regulatory responsibility with private industry would promote efficiency.¹⁴ Google does not allow apps that contain false or misleading health claims,¹⁵ and Apple may scrutinize medical apps that could provide inaccurate data or be used for diagnosing or treating patients.¹⁶ Xcertia, a nonprofit organization founded by the American Medical Association and others, develops standards for the security, privacy, content, and operability of health-related apps, but those standards have not been adopted by other parties. Ultimately, nongovernment agencies are not responsible for public health and do not boast the government’s ability to enforce rules or ensure public safety.

Final Thoughts

The AAD survey of US dermatology residents found that the top consideration when choosing apps was up-to-date and accurate information; however, the 3 most prevalent apps among those same respondents did not need government approval and are not required to contain up-to-date data or to improve clinical outcomes, similar to most other health-related apps. This discrepancy is concerning considering the increasing utilization of apps for physician education and health care delivery and the increasing complexity of those apps. In light of these results, the potential decrease in federal premarket regulation suggested by the FDA’s precertification program seems inappropriate. It is important for the government to take responsibility for regulating health-related apps and to find a balance between too much regulation delaying innovation and too little regulation hurting physician training and patient care. It also is important for providers to be aware of the evidence and oversight behind the technologies they use for professional purposes.

Since the first mobile application (app) was developed in the 1990s, apps have become increasingly integrated into medical practice and training. More than 5.5 million apps were downloadable in 2019,¹ of which more than 300,000 were health related.² In the United States, more than 80% of physicians reported using smartphones for professional purposes in 2016.³ As the complexity of apps and their purpose of use has evolved, regulatory bodies have not adapted adequately to monitor apps that have broad-reaching consequences in medicine.

We review the primary literature on PubMed behind health-related apps that impact dermatologists as well as the government regulation of these apps, with a focus on the 3 most prevalent dermatology-related apps used by dermatology residents in the United States: VisualDx, UpToDate, and Mohs Surgery Appropriate Use Criteria. This prevalence is according to a survey emailed to all dermatology residents in the United States by the American Academy of Dermatology (AAD) in 2019 (unpublished data).

VisualDx

VisualDx, which aims to improve diagnostic accuracy and patient safety, contains peer-reviewed data and more than 32,000 images of dermatologic conditions. The editorial board includes more than 50 physicians. It provides opportunities for continuing medical education credit, is used in more than 2300 medical settings, and costs $399.99 annually for a subscription with partial features. Prior to the launch of the app in 2010, some health science professionals noted that the website version lacked references to primary sources.⁴ The same issue carried over to the app, which has evolved to offer artificial intelligence (AI) analysis of photographed skin lesions. However, there are no peer-reviewed publications showing positive impact of the app on diagnostic skills among dermatology residents or on patient outcomes.

UpToDate

UpToDate is a web-based database created in the early 1990s. A corresponding app was created around 2010. Both internal and independent research has demonstrated improved outcomes, and the app is advertised as the only clinical decision support resource associated with improved outcomes, as shown in more than 80 publications.⁵ UpToDate covers more than 11,800 medical topics and contains more than 35,000 graphics. It cites primary sources and uses a published system for grading recommendation strength and evidence quality. The data are processed and produced by a team of more than 7100 physicians as authors, editors, and reviewers. The platform grants continuing medical education credit and is used by more than 1.9 million clinicians in more than 190 countries. A 1-year subscription for an individual US-based physician costs $559. An observational study assessed UpToDate articles for potential conflicts of interest between authors and their recommendations. Of the 6 articles that met inclusion criteria of discussing management of medical conditions that have controversial or mostly brand-name treatment options, all had conflicts of interest, such as naming drugs from companies with which the authors and/or editors had financial relationships.⁶

Mohs Surgery Appropriate Use Criteria

The Mohs Surgery Appropriate Use Criteria app is a free clinical decision-making tool based on a consensus statement published in 2012 by the AAD, American College of Mohs Surgery, American Society for Dermatologic Surgery Association, and American Society for Mohs Surgery.⁷ It helps guide management of more than 200 dermatologic scenarios. Critique has been made that the criteria are partly based on expert opinion and data largely from the United States and has not been revised to incorporate newer data.⁸ There are no publications regarding the app itself.

Regulation of Health-Related Apps

Health-related apps that are designed for utilization by health care providers can be a valuable tool. However, given their prevalence, cost, and potential impact on patient lives, these apps should be well regulated and researched. The general paucity of peer-reviewed literature demonstrating the utility, safety, quality, and accuracy of health-related apps commonly used by providers is a reflection of insufficient mobile health regulation in the United States.

There are 3 primary government agencies responsible for regulating mobile medical apps: the US Food and Drug Administration (FDA), Federal Trade Commission, and Office for Civil Rights.⁹ The FDA does not regulate all medical devices. Apps intended for use in the diagnosis, cure, mitigation, prevention, or treatment of a disease or condition are considered to be medical devices.¹⁰ The FDA regulates those apps only if they are judged to pose more than minimal risk. Apps that are designed only to provide easy access to information related to health conditions or treatment are considered to be minimal risk but can develop into a different risk level such as by offering AI.¹¹ Although the FDA does update its approach to medical devices, including apps and AI- and machine learning–based software, the rate and direction of update has not kept pace with the rapid evolution of apps.¹² In 2019, the FDA began piloting a precertification program that grants long-term approval to organizations that develop apps instead of reviewing each app product individually.¹³ This decrease in premarket oversight is intended to expedite innovation with the hopeful upside of improving patient outcomes but is inconsistent, with the FDA still reviewing other types of medical devices individually.

For apps that are already in use, the Federal Trade Commission only gets involved in response to deceptive or unfair acts or practices relating to privacy, data security, and false or misleading claims about safety or performance. It may be more beneficial for consumers if those apps had a more stringent initial approval process. The Office for Civil Rights enforces the Health Insurance Portability and Accountability Act when relevant to apps.

Nongovernment agencies also are involved in app regulation. The FDA believes sharing more regulatory responsibility with private industry would promote efficiency.¹⁴ Google does not allow apps that contain false or misleading health claims,¹⁵ and Apple may scrutinize medical apps that could provide inaccurate data or be used for diagnosing or treating patients.¹⁶ Xcertia, a nonprofit organization founded by the American Medical Association and others, develops standards for the security, privacy, content, and operability of health-related apps, but those standards have not been adopted by other parties. Ultimately, nongovernment agencies are not responsible for public health and do not boast the government’s ability to enforce rules or ensure public safety.

Final Thoughts

The AAD survey of US dermatology residents found that the top consideration when choosing apps was up-to-date and accurate information; however, the 3 most prevalent apps among those same respondents did not need government approval and are not required to contain up-to-date data or to improve clinical outcomes, similar to most other health-related apps. This discrepancy is concerning considering the increasing utilization of apps for physician education and health care delivery and the increasing complexity of those apps. In light of these results, the potential decrease in federal premarket regulation suggested by the FDA’s precertification program seems inappropriate. It is important for the government to take responsibility for regulating health-related apps and to find a balance between too much regulation delaying innovation and too little regulation hurting physician training and patient care. It also is important for providers to be aware of the evidence and oversight behind the technologies they use for professional purposes.

Since the first mobile application (app) was developed in the 1990s, apps have become increasingly integrated into medical practice and training. More than 5.5 million apps were downloadable in 2019,¹ of which more than 300,000 were health related.² In the United States, more than 80% of physicians reported using smartphones for professional purposes in 2016.³ As the complexity of apps and their purpose of use has evolved, regulatory bodies have not adapted adequately to monitor apps that have broad-reaching consequences in medicine.

We review the primary literature on PubMed behind health-related apps that impact dermatologists as well as the government regulation of these apps, with a focus on the 3 most prevalent dermatology-related apps used by dermatology residents in the United States: VisualDx, UpToDate, and Mohs Surgery Appropriate Use Criteria. This prevalence is according to a survey emailed to all dermatology residents in the United States by the American Academy of Dermatology (AAD) in 2019 (unpublished data).

VisualDx

VisualDx, which aims to improve diagnostic accuracy and patient safety, contains peer-reviewed data and more than 32,000 images of dermatologic conditions. The editorial board includes more than 50 physicians. It provides opportunities for continuing medical education credit, is used in more than 2300 medical settings, and costs $399.99 annually for a subscription with partial features. Prior to the launch of the app in 2010, some health science professionals noted that the website version lacked references to primary sources.⁴ The same issue carried over to the app, which has evolved to offer artificial intelligence (AI) analysis of photographed skin lesions. However, there are no peer-reviewed publications showing positive impact of the app on diagnostic skills among dermatology residents or on patient outcomes.

UpToDate

UpToDate is a web-based database created in the early 1990s. A corresponding app was created around 2010. Both internal and independent research has demonstrated improved outcomes, and the app is advertised as the only clinical decision support resource associated with improved outcomes, as shown in more than 80 publications.⁵ UpToDate covers more than 11,800 medical topics and contains more than 35,000 graphics. It cites primary sources and uses a published system for grading recommendation strength and evidence quality. The data are processed and produced by a team of more than 7100 physicians as authors, editors, and reviewers. The platform grants continuing medical education credit and is used by more than 1.9 million clinicians in more than 190 countries. A 1-year subscription for an individual US-based physician costs $559. An observational study assessed UpToDate articles for potential conflicts of interest between authors and their recommendations. Of the 6 articles that met inclusion criteria of discussing management of medical conditions that have controversial or mostly brand-name treatment options, all had conflicts of interest, such as naming drugs from companies with which the authors and/or editors had financial relationships.⁶

Mohs Surgery Appropriate Use Criteria

The Mohs Surgery Appropriate Use Criteria app is a free clinical decision-making tool based on a consensus statement published in 2012 by the AAD, American College of Mohs Surgery, American Society for Dermatologic Surgery Association, and American Society for Mohs Surgery.⁷ It helps guide management of more than 200 dermatologic scenarios. Critique has been made that the criteria are partly based on expert opinion and data largely from the United States and has not been revised to incorporate newer data.⁸ There are no publications regarding the app itself.

Regulation of Health-Related Apps

Health-related apps that are designed for utilization by health care providers can be a valuable tool. However, given their prevalence, cost, and potential impact on patient lives, these apps should be well regulated and researched. The general paucity of peer-reviewed literature demonstrating the utility, safety, quality, and accuracy of health-related apps commonly used by providers is a reflection of insufficient mobile health regulation in the United States.

There are 3 primary government agencies responsible for regulating mobile medical apps: the US Food and Drug Administration (FDA), Federal Trade Commission, and Office for Civil Rights.⁹ The FDA does not regulate all medical devices. Apps intended for use in the diagnosis, cure, mitigation, prevention, or treatment of a disease or condition are considered to be medical devices.¹⁰ The FDA regulates those apps only if they are judged to pose more than minimal risk. Apps that are designed only to provide easy access to information related to health conditions or treatment are considered to be minimal risk but can develop into a different risk level such as by offering AI.¹¹ Although the FDA does update its approach to medical devices, including apps and AI- and machine learning–based software, the rate and direction of update has not kept pace with the rapid evolution of apps.¹² In 2019, the FDA began piloting a precertification program that grants long-term approval to organizations that develop apps instead of reviewing each app product individually.¹³ This decrease in premarket oversight is intended to expedite innovation with the hopeful upside of improving patient outcomes but is inconsistent, with the FDA still reviewing other types of medical devices individually.

For apps that are already in use, the Federal Trade Commission only gets involved in response to deceptive or unfair acts or practices relating to privacy, data security, and false or misleading claims about safety or performance. It may be more beneficial for consumers if those apps had a more stringent initial approval process. The Office for Civil Rights enforces the Health Insurance Portability and Accountability Act when relevant to apps.

Nongovernment agencies also are involved in app regulation. The FDA believes sharing more regulatory responsibility with private industry would promote efficiency.¹⁴ Google does not allow apps that contain false or misleading health claims,¹⁵ and Apple may scrutinize medical apps that could provide inaccurate data or be used for diagnosing or treating patients.¹⁶ Xcertia, a nonprofit organization founded by the American Medical Association and others, develops standards for the security, privacy, content, and operability of health-related apps, but those standards have not been adopted by other parties. Ultimately, nongovernment agencies are not responsible for public health and do not boast the government’s ability to enforce rules or ensure public safety.

Final Thoughts

The AAD survey of US dermatology residents found that the top consideration when choosing apps was up-to-date and accurate information; however, the 3 most prevalent apps among those same respondents did not need government approval and are not required to contain up-to-date data or to improve clinical outcomes, similar to most other health-related apps. This discrepancy is concerning considering the increasing utilization of apps for physician education and health care delivery and the increasing complexity of those apps. In light of these results, the potential decrease in federal premarket regulation suggested by the FDA’s precertification program seems inappropriate. It is important for the government to take responsibility for regulating health-related apps and to find a balance between too much regulation delaying innovation and too little regulation hurting physician training and patient care. It also is important for providers to be aware of the evidence and oversight behind the technologies they use for professional purposes.

Artificial intelligence (AI) is a loosely defined term that refers to machines (ie, algorithms) simulating facets of human intelligence. Some examples of AI are seen in natural language-processing algorithms, including autocorrect and search engine autocomplete functions; voice recognition in virtual assistants; autopilot systems in airplanes and self-driving cars; and computer vision in image and object recognition. Since the dawn of the century, various forms of AI have been tested and introduced in health care. However, a gap exists between clinician viewpoints on AI and the engineering world’s assumptions of what can be automated in medicine.

In this article, we review the history and evolution of AI in medicine, focusing on radiology and dermatology; current capabilities of AI; challenges to clinical integration; and future directions. Our aim is to provide realistic expectations of current technologies in solving complex problems and to empower dermatologists in planning for a future that likely includes various forms of AI.

Early Stages of AI in Medical Decision-making

Some of the earliest forms of clinical decision-support software in medicine were computer-aided detection and computer-aided diagnosis (CAD) used in screening for breast and lung cancer on mammography and computed tomography.^1-3 Early research on the use of CAD systems in radiology date to the 1960s (Figure), with the first US Food and Drug Administration–approved CAD system in mammography in 1998 and for Centers for Medicare & Medicaid Services reimbursement in 2002.^1,2

Early CAD systems relied on rule-based classifiers, which use predefined features to classify images into desired categories. For example, to classify an image as a high-risk or benign mass, features such as contour and texture had to be explicitly defined. Although these systems showed on par with, or higher, accuracy vs a radiologist in validation studies, early CAD systems never achieved wide adoption because of an increased rate of false positives as well as added work burden on a radiologist, who had to silence overcalling by the software.^1,2,4,5

Computer-aided diagnosis–based melanoma diagnosis was introduced in early 2000 in dermatology (Figure) using the same feature-based classifiers. These systems claimed expert-level accuracy in proof-of-concept studies and prospective uncontrolled trials on proprietary devices using these classifiers.^6,7 Similar to radiology, however, real-world adoption did not happen; in fact, the last of these devices was taken off the market in 2017. A recent meta-analysis of studies using CAD-based melanoma diagnosis point to study bias; data overfitting; and lack of large controlled, prospective trials as possible reasons why results could not be replicated in a clinical setting.⁸

Beyond 2010: Deep Learning

New techniques in machine learning (ML), called deep learning, began to emerge after 2010 (Figure). In deep learning, instead of directing the computer to look for certain discriminative features, the machine learns those features from the large amount of data without being explicitly programed to do so. In other words, compared to predecessor forms of computing, there is less human supervision in the learning process (Table). The concept of ML has existed since the 1980s. The field saw exponential growth in the last decade with the improvement of algorithms; an increase in computing power; and emergence of large training data sets, such as open-source platforms on the Web.^9,10

Most ML methods today incorporate artificial neural networks (ANN), computer programs that imitate the architecture of biological neural networks and form dynamically changing systems that improve with continuous data exposure. The performance of an ANN is dependent on the number and architecture of its neural layers and (similar to CAD systems) the size, quality, and generalizability of the training data set.^9-12

In medicine, images (eg, clinical or dermoscopic images and imaging scans) are the most commonly used form of data for AI development. Convolutional neural networks (CNN), a subtype of ANN, are frequently used for this purpose. These networks use a hierarchical neural network architecture, similar to the visual cortex, that allows for composition of complex features (eg, shapes) from simpler features (eg, image intensities), which leads to more efficient data processing.^10-12

In recent years, CNNs have been applied in a number of image-based medical fields, including radiology, dermatology, and pathology. Initially, studies were largely led by computer scientists trying to match clinician performance in detection of disease categories. However, there has been a shift toward more physicians getting involved, which has motivated development of large curated (ie, expert-labeled) and standardized clinical data sets in training the CNN. Although training on quality-controlled data is a work in progress across medical disciplines, it has led to improved machine performance.^11,12

Recent Advances in AI

In recent years, the number of studies covering CNN in diagnosis has increased exponentially in several medical specialties. The goal is to improve software to close the gap between experts and the machine in live clinical settings. The current literature focuses on a comparison of experts with the machine in simulated settings; prospective clinical trials are still lagging in the real world.^9,11,13

We look at radiology to explore recent advances in AI diagnosis for 3 reasons: (1) radiology has the largest repository of digital data (using a picture archiving and communication system) among medical specialties; (2) radiology has well-defined, image-acquisition protocols in its clinical workflow¹⁴; and (3) gray-scale images are easier to standardize because they are impervious to environmental variables that are difficult to control (eg, recent sun exposure, rosacea flare, lighting, sweating). These are some of the reasons we think radiology is, and will be, ahead in training AI algorithms and integrating them into clinical practice. However, even radiology AI studies have limitations, including a lack of prospective, real-world clinical setting, generalizable studies, and a lack of large standardized available databases for training algorithms.

Narrowing our discussion to studies of mammography—given the repetitive nature and binary output of this modality, which has made it one of the first targets of automation in diagnostic imaging^1,2,5,13—AI-based CAD in mammography, much like its predecessor feature-based CAD, has shown promising results in artificial settings. Five key mammography CNN studies have reported a wide range of diagnostic accuracy (area under the curve, 69.2 to 97.8 [mean, 88.2]) compared to radiologists.^15-19

In the most recent study (2019), Rodriguez-Ruiz et al¹⁵ compared machines and a cohort of 101 radiologists, in which AI showed performance comparability. However, results in this artificial setting were not followed up with prospective analysis of the technology in a clinical setting. First-generation, feature-based CADs in mammography also showed expert-level performance in artificial settings, but the technology became extinct because these results were not generalizable to real-world in prospective trials. To our knowledge, a limitation of radiology AI is that all current CNNs have not yet been tested in a live clinical setting.^13-19

The second limitation of radiology AI is lack of standardization, which also applies to mammography, despite this subset having the largest and oldest publicly available data set. In a recent review of 23 studies on AI-based algorithms in mammography (2010-2019), clinicians point to one of the biggest flaws: the use of small, nonstandardized, and skewed public databases (often enriched for malignancy) as training algorithms.¹³

Standardization refers to quality-control measures in acquisition, processing, and image labeling that need to be met for images to be included in the training data set. At present, large stores of radiologic data that are standardized within each institution are not publicly accessible through a unified reference platform. Lack of large standardized training data sets leads to selection bias and increases the risk for overfitting, which occurs when algorithm models incorporate background noise in the data into its prediction scheme. Overfitting has been noted in several AI-based studies in mammography,¹³ which limits the generalizability of algorithm performance in the real-world setting.

To overcome this limitation, the American College of Radiology Data Science Institute recently took the lead on creating a reference platform for quality control and standardized data generation for AI integration in radiology. The goal of the institute is for radiologists to work collaboratively with industry to ensure that algorithms are trained on quality data that produces clinically useable output for the clinician and patient.^11,20

Similar to initial radiology studies utilizing AI mainly as a screening tool, AI-driven studies in dermatology are focused on classification of melanocytic lesions; the goal is to aid in melanoma screening. Two of the most-recent, most-cited articles on this topic are by Esteva et al²¹ and Tschandl et al.²² Esteva et al²¹ matched the performance of 21 dermatologists in binary classification (malignant or nonmalignant) of clinical and dermoscopic images in pigmented and nonpigmented categories. A CNN developed by Google was trained on 130,000 clinical images encompassing more than 2000 dermatologist-labeled diagnoses from 18 sites. Despite promising results, the question remains whether these findings are transferrable to the clinical setting. In addition to the limitation on generalizability, the authors do not elaborate on standardization of training image data sets. For example, it is unclear what percentage of the training data set’s image labels were based on biopsy results vs clinical diagnosis.²¹

The second study was the largest Web-based study to compare the performance of more than 500 dermatologists worldwide.²² The top 3–performing algorithms (among a pool of 139) were at least as good as the performance of 27 expert dermatologists (defined as having more than 10 years’ experience) in the classification of pigmented lesions into 7 predefined categories.²² However, images came from nonstandardized sources gathered from a 20-year period at one European academic center and a private practice in Australia. Tschandl et al²² looked at external validation with an independent data set, outside the training data set. Although not generalizable to a real-world setting, looking at external data sets helps correct for overfitting and is a good first step in understanding transferability of results. However, the external data set was chosen by the authors and therefore might be tainted by selection bias. Although only a 10% drop in algorithmic accuracy was noted using the external data set chosen by the authors, this drop does not apply to other data sets or more importantly to a real-world setting.²²

Current limitations and future goals of radiology also will most likely apply to dermatology AI research. In medicine and radiology, the goal of AI is to first help users by prioritizing what they should focus on. The concept of comparing AI to a radiologist or dermatologist is potentially shortsighted. Shortcomings of the current supervised or semisupervised algorithms used in medicine underscore the points that, first, to make their outputs clinically usable, it should be clinicians who procure and standardize training data sets and, second, it appears logical that the performance of these category of algorithms requires constant monitoring for bias. Therefore, these algorithms cannot operate as stand-alone diagnostic machines but as an aid to the clinician—if the performance of the algorithms is proved in large trials.

Near-Future Directions and Projections

Almost all recent state-of-the-art AI systems tested in medical disciplines fall under the engineering terminology of narrow or weak AI, meaning any given algorithm is trained to do only one specific task.⁹ An example of a task is classification of images into multiple categories (ie, benign or malignant). However, task classification only works with preselected images that will need substantial improvements in standardization.

Although it has been demonstrated that AI systems can excel at one task at a time, such as classification, better than a human cohort in simulated settings, these literal machines lack the ability to incorporate context; integrate various forms of sensory input such as visual, voice, or text; or make associations the way humans do.⁹ Multiple tasks and clinical context integration are required for predictive diagnosis or clinical decision-making, even in a simulated environment. In this sense, CNN is still similar to its antiquated linear CAD predecessor: It cannot make a diagnosis or a clinical decision but might be appropriate for triaging cases that are referred for evaluation by a dermatologist.

Medical AI also may use electronic health records or patient-gathered data (eg, apps). However, clinical images are more structured and less noisy and are more easily incorporated in AI training. Therefore, as we are already witnessing, earlier validation and adoption of AI will occur in image-based disciplines, beginning with radiology; then pathology; and eventually dermatology, which will be the most challenging of the 3 medical specialties to standardize.

Final Thoughts

Artificial intelligence in health care is in its infancy; specific task-driven algorithms are only beginning to be introduced. We project that in the next 5 to 10 years, clinicians will become increasingly involved in training and testing large-scale validation as well as monitoring narrow AI in clinical trials. Radiology has served as the pioneering area in medicine and is just beginning to utilize narrow AI to help specialists with very specific tasks. For example, a task would be to triage which scans to look at first for a radiologist or which pigmented lesion might need prompt evaluation by a dermatologist. Artificial intelligence in medicine is not replacing specialists or placing decision-making in the hands of a nonexpert. At this point, CNNs have not proven that they make us better at diagnosing because real-world clinical data are lacking, which may change in the future with large standardized training data sets and validation with prospective clinical trials. The near future for dermatology and pathology will follow what is already happening in radiology, with AI substantially increasing workflow efficiency by prioritizing tasks.

Artificial intelligence (AI) is a loosely defined term that refers to machines (ie, algorithms) simulating facets of human intelligence. Some examples of AI are seen in natural language-processing algorithms, including autocorrect and search engine autocomplete functions; voice recognition in virtual assistants; autopilot systems in airplanes and self-driving cars; and computer vision in image and object recognition. Since the dawn of the century, various forms of AI have been tested and introduced in health care. However, a gap exists between clinician viewpoints on AI and the engineering world’s assumptions of what can be automated in medicine.

In this article, we review the history and evolution of AI in medicine, focusing on radiology and dermatology; current capabilities of AI; challenges to clinical integration; and future directions. Our aim is to provide realistic expectations of current technologies in solving complex problems and to empower dermatologists in planning for a future that likely includes various forms of AI.

Early Stages of AI in Medical Decision-making

Some of the earliest forms of clinical decision-support software in medicine were computer-aided detection and computer-aided diagnosis (CAD) used in screening for breast and lung cancer on mammography and computed tomography.^1-3 Early research on the use of CAD systems in radiology date to the 1960s (Figure), with the first US Food and Drug Administration–approved CAD system in mammography in 1998 and for Centers for Medicare & Medicaid Services reimbursement in 2002.^1,2

Early CAD systems relied on rule-based classifiers, which use predefined features to classify images into desired categories. For example, to classify an image as a high-risk or benign mass, features such as contour and texture had to be explicitly defined. Although these systems showed on par with, or higher, accuracy vs a radiologist in validation studies, early CAD systems never achieved wide adoption because of an increased rate of false positives as well as added work burden on a radiologist, who had to silence overcalling by the software.^1,2,4,5

Computer-aided diagnosis–based melanoma diagnosis was introduced in early 2000 in dermatology (Figure) using the same feature-based classifiers. These systems claimed expert-level accuracy in proof-of-concept studies and prospective uncontrolled trials on proprietary devices using these classifiers.^6,7 Similar to radiology, however, real-world adoption did not happen; in fact, the last of these devices was taken off the market in 2017. A recent meta-analysis of studies using CAD-based melanoma diagnosis point to study bias; data overfitting; and lack of large controlled, prospective trials as possible reasons why results could not be replicated in a clinical setting.⁸

Beyond 2010: Deep Learning

New techniques in machine learning (ML), called deep learning, began to emerge after 2010 (Figure). In deep learning, instead of directing the computer to look for certain discriminative features, the machine learns those features from the large amount of data without being explicitly programed to do so. In other words, compared to predecessor forms of computing, there is less human supervision in the learning process (Table). The concept of ML has existed since the 1980s. The field saw exponential growth in the last decade with the improvement of algorithms; an increase in computing power; and emergence of large training data sets, such as open-source platforms on the Web.^9,10

Most ML methods today incorporate artificial neural networks (ANN), computer programs that imitate the architecture of biological neural networks and form dynamically changing systems that improve with continuous data exposure. The performance of an ANN is dependent on the number and architecture of its neural layers and (similar to CAD systems) the size, quality, and generalizability of the training data set.^9-12

In medicine, images (eg, clinical or dermoscopic images and imaging scans) are the most commonly used form of data for AI development. Convolutional neural networks (CNN), a subtype of ANN, are frequently used for this purpose. These networks use a hierarchical neural network architecture, similar to the visual cortex, that allows for composition of complex features (eg, shapes) from simpler features (eg, image intensities), which leads to more efficient data processing.^10-12

In recent years, CNNs have been applied in a number of image-based medical fields, including radiology, dermatology, and pathology. Initially, studies were largely led by computer scientists trying to match clinician performance in detection of disease categories. However, there has been a shift toward more physicians getting involved, which has motivated development of large curated (ie, expert-labeled) and standardized clinical data sets in training the CNN. Although training on quality-controlled data is a work in progress across medical disciplines, it has led to improved machine performance.^11,12

Recent Advances in AI

In recent years, the number of studies covering CNN in diagnosis has increased exponentially in several medical specialties. The goal is to improve software to close the gap between experts and the machine in live clinical settings. The current literature focuses on a comparison of experts with the machine in simulated settings; prospective clinical trials are still lagging in the real world.^9,11,13

We look at radiology to explore recent advances in AI diagnosis for 3 reasons: (1) radiology has the largest repository of digital data (using a picture archiving and communication system) among medical specialties; (2) radiology has well-defined, image-acquisition protocols in its clinical workflow¹⁴; and (3) gray-scale images are easier to standardize because they are impervious to environmental variables that are difficult to control (eg, recent sun exposure, rosacea flare, lighting, sweating). These are some of the reasons we think radiology is, and will be, ahead in training AI algorithms and integrating them into clinical practice. However, even radiology AI studies have limitations, including a lack of prospective, real-world clinical setting, generalizable studies, and a lack of large standardized available databases for training algorithms.

Narrowing our discussion to studies of mammography—given the repetitive nature and binary output of this modality, which has made it one of the first targets of automation in diagnostic imaging^1,2,5,13—AI-based CAD in mammography, much like its predecessor feature-based CAD, has shown promising results in artificial settings. Five key mammography CNN studies have reported a wide range of diagnostic accuracy (area under the curve, 69.2 to 97.8 [mean, 88.2]) compared to radiologists.^15-19

In the most recent study (2019), Rodriguez-Ruiz et al¹⁵ compared machines and a cohort of 101 radiologists, in which AI showed performance comparability. However, results in this artificial setting were not followed up with prospective analysis of the technology in a clinical setting. First-generation, feature-based CADs in mammography also showed expert-level performance in artificial settings, but the technology became extinct because these results were not generalizable to real-world in prospective trials. To our knowledge, a limitation of radiology AI is that all current CNNs have not yet been tested in a live clinical setting.^13-19

The second limitation of radiology AI is lack of standardization, which also applies to mammography, despite this subset having the largest and oldest publicly available data set. In a recent review of 23 studies on AI-based algorithms in mammography (2010-2019), clinicians point to one of the biggest flaws: the use of small, nonstandardized, and skewed public databases (often enriched for malignancy) as training algorithms.¹³

Standardization refers to quality-control measures in acquisition, processing, and image labeling that need to be met for images to be included in the training data set. At present, large stores of radiologic data that are standardized within each institution are not publicly accessible through a unified reference platform. Lack of large standardized training data sets leads to selection bias and increases the risk for overfitting, which occurs when algorithm models incorporate background noise in the data into its prediction scheme. Overfitting has been noted in several AI-based studies in mammography,¹³ which limits the generalizability of algorithm performance in the real-world setting.

To overcome this limitation, the American College of Radiology Data Science Institute recently took the lead on creating a reference platform for quality control and standardized data generation for AI integration in radiology. The goal of the institute is for radiologists to work collaboratively with industry to ensure that algorithms are trained on quality data that produces clinically useable output for the clinician and patient.^11,20

Similar to initial radiology studies utilizing AI mainly as a screening tool, AI-driven studies in dermatology are focused on classification of melanocytic lesions; the goal is to aid in melanoma screening. Two of the most-recent, most-cited articles on this topic are by Esteva et al²¹ and Tschandl et al.²² Esteva et al²¹ matched the performance of 21 dermatologists in binary classification (malignant or nonmalignant) of clinical and dermoscopic images in pigmented and nonpigmented categories. A CNN developed by Google was trained on 130,000 clinical images encompassing more than 2000 dermatologist-labeled diagnoses from 18 sites. Despite promising results, the question remains whether these findings are transferrable to the clinical setting. In addition to the limitation on generalizability, the authors do not elaborate on standardization of training image data sets. For example, it is unclear what percentage of the training data set’s image labels were based on biopsy results vs clinical diagnosis.²¹

The second study was the largest Web-based study to compare the performance of more than 500 dermatologists worldwide.²² The top 3–performing algorithms (among a pool of 139) were at least as good as the performance of 27 expert dermatologists (defined as having more than 10 years’ experience) in the classification of pigmented lesions into 7 predefined categories.²² However, images came from nonstandardized sources gathered from a 20-year period at one European academic center and a private practice in Australia. Tschandl et al²² looked at external validation with an independent data set, outside the training data set. Although not generalizable to a real-world setting, looking at external data sets helps correct for overfitting and is a good first step in understanding transferability of results. However, the external data set was chosen by the authors and therefore might be tainted by selection bias. Although only a 10% drop in algorithmic accuracy was noted using the external data set chosen by the authors, this drop does not apply to other data sets or more importantly to a real-world setting.²²

Current limitations and future goals of radiology also will most likely apply to dermatology AI research. In medicine and radiology, the goal of AI is to first help users by prioritizing what they should focus on. The concept of comparing AI to a radiologist or dermatologist is potentially shortsighted. Shortcomings of the current supervised or semisupervised algorithms used in medicine underscore the points that, first, to make their outputs clinically usable, it should be clinicians who procure and standardize training data sets and, second, it appears logical that the performance of these category of algorithms requires constant monitoring for bias. Therefore, these algorithms cannot operate as stand-alone diagnostic machines but as an aid to the clinician—if the performance of the algorithms is proved in large trials.

Near-Future Directions and Projections

Almost all recent state-of-the-art AI systems tested in medical disciplines fall under the engineering terminology of narrow or weak AI, meaning any given algorithm is trained to do only one specific task.⁹ An example of a task is classification of images into multiple categories (ie, benign or malignant). However, task classification only works with preselected images that will need substantial improvements in standardization.

Although it has been demonstrated that AI systems can excel at one task at a time, such as classification, better than a human cohort in simulated settings, these literal machines lack the ability to incorporate context; integrate various forms of sensory input such as visual, voice, or text; or make associations the way humans do.⁹ Multiple tasks and clinical context integration are required for predictive diagnosis or clinical decision-making, even in a simulated environment. In this sense, CNN is still similar to its antiquated linear CAD predecessor: It cannot make a diagnosis or a clinical decision but might be appropriate for triaging cases that are referred for evaluation by a dermatologist.

Medical AI also may use electronic health records or patient-gathered data (eg, apps). However, clinical images are more structured and less noisy and are more easily incorporated in AI training. Therefore, as we are already witnessing, earlier validation and adoption of AI will occur in image-based disciplines, beginning with radiology; then pathology; and eventually dermatology, which will be the most challenging of the 3 medical specialties to standardize.

Final Thoughts

Artificial intelligence in health care is in its infancy; specific task-driven algorithms are only beginning to be introduced. We project that in the next 5 to 10 years, clinicians will become increasingly involved in training and testing large-scale validation as well as monitoring narrow AI in clinical trials. Radiology has served as the pioneering area in medicine and is just beginning to utilize narrow AI to help specialists with very specific tasks. For example, a task would be to triage which scans to look at first for a radiologist or which pigmented lesion might need prompt evaluation by a dermatologist. Artificial intelligence in medicine is not replacing specialists or placing decision-making in the hands of a nonexpert. At this point, CNNs have not proven that they make us better at diagnosing because real-world clinical data are lacking, which may change in the future with large standardized training data sets and validation with prospective clinical trials. The near future for dermatology and pathology will follow what is already happening in radiology, with AI substantially increasing workflow efficiency by prioritizing tasks.

Artificial intelligence (AI) is a loosely defined term that refers to machines (ie, algorithms) simulating facets of human intelligence. Some examples of AI are seen in natural language-processing algorithms, including autocorrect and search engine autocomplete functions; voice recognition in virtual assistants; autopilot systems in airplanes and self-driving cars; and computer vision in image and object recognition. Since the dawn of the century, various forms of AI have been tested and introduced in health care. However, a gap exists between clinician viewpoints on AI and the engineering world’s assumptions of what can be automated in medicine.

In this article, we review the history and evolution of AI in medicine, focusing on radiology and dermatology; current capabilities of AI; challenges to clinical integration; and future directions. Our aim is to provide realistic expectations of current technologies in solving complex problems and to empower dermatologists in planning for a future that likely includes various forms of AI.

Early Stages of AI in Medical Decision-making

Some of the earliest forms of clinical decision-support software in medicine were computer-aided detection and computer-aided diagnosis (CAD) used in screening for breast and lung cancer on mammography and computed tomography.^1-3 Early research on the use of CAD systems in radiology date to the 1960s (Figure), with the first US Food and Drug Administration–approved CAD system in mammography in 1998 and for Centers for Medicare & Medicaid Services reimbursement in 2002.^1,2

Early CAD systems relied on rule-based classifiers, which use predefined features to classify images into desired categories. For example, to classify an image as a high-risk or benign mass, features such as contour and texture had to be explicitly defined. Although these systems showed on par with, or higher, accuracy vs a radiologist in validation studies, early CAD systems never achieved wide adoption because of an increased rate of false positives as well as added work burden on a radiologist, who had to silence overcalling by the software.^1,2,4,5

Computer-aided diagnosis–based melanoma diagnosis was introduced in early 2000 in dermatology (Figure) using the same feature-based classifiers. These systems claimed expert-level accuracy in proof-of-concept studies and prospective uncontrolled trials on proprietary devices using these classifiers.^6,7 Similar to radiology, however, real-world adoption did not happen; in fact, the last of these devices was taken off the market in 2017. A recent meta-analysis of studies using CAD-based melanoma diagnosis point to study bias; data overfitting; and lack of large controlled, prospective trials as possible reasons why results could not be replicated in a clinical setting.⁸

Beyond 2010: Deep Learning

New techniques in machine learning (ML), called deep learning, began to emerge after 2010 (Figure). In deep learning, instead of directing the computer to look for certain discriminative features, the machine learns those features from the large amount of data without being explicitly programed to do so. In other words, compared to predecessor forms of computing, there is less human supervision in the learning process (Table). The concept of ML has existed since the 1980s. The field saw exponential growth in the last decade with the improvement of algorithms; an increase in computing power; and emergence of large training data sets, such as open-source platforms on the Web.^9,10

Most ML methods today incorporate artificial neural networks (ANN), computer programs that imitate the architecture of biological neural networks and form dynamically changing systems that improve with continuous data exposure. The performance of an ANN is dependent on the number and architecture of its neural layers and (similar to CAD systems) the size, quality, and generalizability of the training data set.^9-12

In medicine, images (eg, clinical or dermoscopic images and imaging scans) are the most commonly used form of data for AI development. Convolutional neural networks (CNN), a subtype of ANN, are frequently used for this purpose. These networks use a hierarchical neural network architecture, similar to the visual cortex, that allows for composition of complex features (eg, shapes) from simpler features (eg, image intensities), which leads to more efficient data processing.^10-12

In recent years, CNNs have been applied in a number of image-based medical fields, including radiology, dermatology, and pathology. Initially, studies were largely led by computer scientists trying to match clinician performance in detection of disease categories. However, there has been a shift toward more physicians getting involved, which has motivated development of large curated (ie, expert-labeled) and standardized clinical data sets in training the CNN. Although training on quality-controlled data is a work in progress across medical disciplines, it has led to improved machine performance.^11,12

Recent Advances in AI

In recent years, the number of studies covering CNN in diagnosis has increased exponentially in several medical specialties. The goal is to improve software to close the gap between experts and the machine in live clinical settings. The current literature focuses on a comparison of experts with the machine in simulated settings; prospective clinical trials are still lagging in the real world.^9,11,13

We look at radiology to explore recent advances in AI diagnosis for 3 reasons: (1) radiology has the largest repository of digital data (using a picture archiving and communication system) among medical specialties; (2) radiology has well-defined, image-acquisition protocols in its clinical workflow¹⁴; and (3) gray-scale images are easier to standardize because they are impervious to environmental variables that are difficult to control (eg, recent sun exposure, rosacea flare, lighting, sweating). These are some of the reasons we think radiology is, and will be, ahead in training AI algorithms and integrating them into clinical practice. However, even radiology AI studies have limitations, including a lack of prospective, real-world clinical setting, generalizable studies, and a lack of large standardized available databases for training algorithms.

Narrowing our discussion to studies of mammography—given the repetitive nature and binary output of this modality, which has made it one of the first targets of automation in diagnostic imaging^1,2,5,13—AI-based CAD in mammography, much like its predecessor feature-based CAD, has shown promising results in artificial settings. Five key mammography CNN studies have reported a wide range of diagnostic accuracy (area under the curve, 69.2 to 97.8 [mean, 88.2]) compared to radiologists.^15-19

In the most recent study (2019), Rodriguez-Ruiz et al¹⁵ compared machines and a cohort of 101 radiologists, in which AI showed performance comparability. However, results in this artificial setting were not followed up with prospective analysis of the technology in a clinical setting. First-generation, feature-based CADs in mammography also showed expert-level performance in artificial settings, but the technology became extinct because these results were not generalizable to real-world in prospective trials. To our knowledge, a limitation of radiology AI is that all current CNNs have not yet been tested in a live clinical setting.^13-19

The second limitation of radiology AI is lack of standardization, which also applies to mammography, despite this subset having the largest and oldest publicly available data set. In a recent review of 23 studies on AI-based algorithms in mammography (2010-2019), clinicians point to one of the biggest flaws: the use of small, nonstandardized, and skewed public databases (often enriched for malignancy) as training algorithms.¹³

Standardization refers to quality-control measures in acquisition, processing, and image labeling that need to be met for images to be included in the training data set. At present, large stores of radiologic data that are standardized within each institution are not publicly accessible through a unified reference platform. Lack of large standardized training data sets leads to selection bias and increases the risk for overfitting, which occurs when algorithm models incorporate background noise in the data into its prediction scheme. Overfitting has been noted in several AI-based studies in mammography,¹³ which limits the generalizability of algorithm performance in the real-world setting.

To overcome this limitation, the American College of Radiology Data Science Institute recently took the lead on creating a reference platform for quality control and standardized data generation for AI integration in radiology. The goal of the institute is for radiologists to work collaboratively with industry to ensure that algorithms are trained on quality data that produces clinically useable output for the clinician and patient.^11,20

Similar to initial radiology studies utilizing AI mainly as a screening tool, AI-driven studies in dermatology are focused on classification of melanocytic lesions; the goal is to aid in melanoma screening. Two of the most-recent, most-cited articles on this topic are by Esteva et al²¹ and Tschandl et al.²² Esteva et al²¹ matched the performance of 21 dermatologists in binary classification (malignant or nonmalignant) of clinical and dermoscopic images in pigmented and nonpigmented categories. A CNN developed by Google was trained on 130,000 clinical images encompassing more than 2000 dermatologist-labeled diagnoses from 18 sites. Despite promising results, the question remains whether these findings are transferrable to the clinical setting. In addition to the limitation on generalizability, the authors do not elaborate on standardization of training image data sets. For example, it is unclear what percentage of the training data set’s image labels were based on biopsy results vs clinical diagnosis.²¹

The second study was the largest Web-based study to compare the performance of more than 500 dermatologists worldwide.²² The top 3–performing algorithms (among a pool of 139) were at least as good as the performance of 27 expert dermatologists (defined as having more than 10 years’ experience) in the classification of pigmented lesions into 7 predefined categories.²² However, images came from nonstandardized sources gathered from a 20-year period at one European academic center and a private practice in Australia. Tschandl et al²² looked at external validation with an independent data set, outside the training data set. Although not generalizable to a real-world setting, looking at external data sets helps correct for overfitting and is a good first step in understanding transferability of results. However, the external data set was chosen by the authors and therefore might be tainted by selection bias. Although only a 10% drop in algorithmic accuracy was noted using the external data set chosen by the authors, this drop does not apply to other data sets or more importantly to a real-world setting.²²

Current limitations and future goals of radiology also will most likely apply to dermatology AI research. In medicine and radiology, the goal of AI is to first help users by prioritizing what they should focus on. The concept of comparing AI to a radiologist or dermatologist is potentially shortsighted. Shortcomings of the current supervised or semisupervised algorithms used in medicine underscore the points that, first, to make their outputs clinically usable, it should be clinicians who procure and standardize training data sets and, second, it appears logical that the performance of these category of algorithms requires constant monitoring for bias. Therefore, these algorithms cannot operate as stand-alone diagnostic machines but as an aid to the clinician—if the performance of the algorithms is proved in large trials.

Near-Future Directions and Projections

Almost all recent state-of-the-art AI systems tested in medical disciplines fall under the engineering terminology of narrow or weak AI, meaning any given algorithm is trained to do only one specific task.⁹ An example of a task is classification of images into multiple categories (ie, benign or malignant). However, task classification only works with preselected images that will need substantial improvements in standardization.

Although it has been demonstrated that AI systems can excel at one task at a time, such as classification, better than a human cohort in simulated settings, these literal machines lack the ability to incorporate context; integrate various forms of sensory input such as visual, voice, or text; or make associations the way humans do.⁹ Multiple tasks and clinical context integration are required for predictive diagnosis or clinical decision-making, even in a simulated environment. In this sense, CNN is still similar to its antiquated linear CAD predecessor: It cannot make a diagnosis or a clinical decision but might be appropriate for triaging cases that are referred for evaluation by a dermatologist.

Medical AI also may use electronic health records or patient-gathered data (eg, apps). However, clinical images are more structured and less noisy and are more easily incorporated in AI training. Therefore, as we are already witnessing, earlier validation and adoption of AI will occur in image-based disciplines, beginning with radiology; then pathology; and eventually dermatology, which will be the most challenging of the 3 medical specialties to standardize.

Final Thoughts

Artificial intelligence in health care is in its infancy; specific task-driven algorithms are only beginning to be introduced. We project that in the next 5 to 10 years, clinicians will become increasingly involved in training and testing large-scale validation as well as monitoring narrow AI in clinical trials. Radiology has served as the pioneering area in medicine and is just beginning to utilize narrow AI to help specialists with very specific tasks. For example, a task would be to triage which scans to look at first for a radiologist or which pigmented lesion might need prompt evaluation by a dermatologist. Artificial intelligence in medicine is not replacing specialists or placing decision-making in the hands of a nonexpert. At this point, CNNs have not proven that they make us better at diagnosing because real-world clinical data are lacking, which may change in the future with large standardized training data sets and validation with prospective clinical trials. The near future for dermatology and pathology will follow what is already happening in radiology, with AI substantially increasing workflow efficiency by prioritizing tasks.

Traditionally, diagnosis of skin disease relies on clinical inspection, often followed by biopsy and histopathologic examination. In recent years, new noninvasive tools have emerged that can aid in clinical diagnosis and reduce the number of unnecessary benign biopsies. Although there has been a surge in noninvasive diagnostic technologies, many tools are still in research and development phases, with few tools widely adopted and used in regular clinical practice. In this article, we discuss the use of dermoscopy, reflectance confocal microscopy (RCM), and optical coherence tomography (OCT) in the diagnosis and management of skin disease.

Dermoscopy

Dermoscopy, also known as epiluminescence light microscopy and previously known as dermatoscopy, utilizes a ×10 to ×100 microscope objective with a light source to magnify and visualize structures present below the skin’s surface, such as melanin and blood vessels. There are 3 types of dermoscopy: conventional nonpolarized dermoscopy, polarized contact dermoscopy, and nonpolarized contact dermoscopy (Figure 1). Traditional nonpolarized dermoscopy requires a liquid medium and direct contact with the skin, and it relies on light reflection and refraction properties.¹ Cross-polarized light sources allow visualization of deeper structures, either with or without a liquid medium and contact with the skin surface. Although there is overall concurrence among the different types of dermoscopy, subtle differences in the appearance of color, features, and structure are present.¹

Dermoscopy offers many benefits for dermatologists and other providers. It can be used to aid in the diagnosis of cutaneous neoplasms and other skin diseases. Numerous low-cost dermatoscopes currently are commercially available. The handheld, easily transportable nature of dermatoscopes have resulted in widespread practice integration. Approximately 84% of attending dermatologists in US academic settings reported using dermoscopy, and many refer to the dermatoscope as “the dermatologist’s stethoscope.”² In addition, 6% to 15% of other US providers, including family physicians, internal medicine physicians, and plastic surgeons, have reported using dermoscopy in their clinical practices. Limitations of dermoscopy include visualization of the skin surface only and not deeper structures within the tissue, the need for training for adequate interpretation of dermoscopic images, and lack of reimbursement for dermoscopic examination.³

Many dermoscopic structures that correspond well with histopathology have been described. Dermoscopy has a sensitivity of 79% to 96% and specificity of 69% to 99% in the diagnosis of melanoma.⁴ There is variable data on the specificity of dermoscopy in the diagnosis of melanoma, with one meta-analysis finding no statistically significant difference in specificity compared to naked eye examination,⁵ while other studies report increased specificity and subsequent reduction in biopsy of benign lesions.^6,7 Dermoscopy also can aid in the diagnosis of keratinocytic neoplasms, and dermoscopy also results in a sensitivity of 78.6% to 100% and a specificity of 53.8% to 100% in the diagnosis of basal cell carcinoma (BCC).⁸ Limitations of dermoscopy include false-positive diagnoses, commonly seborrheic keratoses and nevi, resulting in unnecessary biopsies, as well as false-negative diagnoses, commonly amelanotic and nevoid melanoma, resulting in delays in skin cancer diagnosis and resultant poor outcomes.⁹ Dermoscopy also is used to aid in the diagnosis of inflammatory and infectious skin diseases, as well as scalp, hair, and nail disorders.¹⁰

Reflectance Confocal Microscopy

Reflectance confocal microscopy utilizes an 830-nm laser to capture horizontal en face images of the skin with high resolution. Different structures of the skin have varying indices of refraction: keratin, melanin, and collagen appear bright white, while other components appear dark, generating black-and-white RCM images.¹¹ Currently, there are 2 reflectance confocal microscopes that are commercially available in the United States. The Vivascope 1500 (Caliber ID) is the traditional model that captures 8×8-mm images, and the Vivascope 3000 (Caliber ID) is a smaller handheld model that captures 0.5×0.5-mm images. The traditional model provides the advantages of higher-resolution images and the ability to capture larger surface areas but is best suited to image flat areas of skin to which a square window can be adhered. The handheld model allows improved contact with the varying topography of skin; does not require an adhesive window; and can be used to image cartilaginous, mucosal, and sensitive surfaces. However, it can be difficult to correlate individual images captured by the handheld RCM with the location relative to the lesion, as it is exquisitely sensitive to motion and also is operator dependent. Although complex algorithms are under development to stitch individual images to provide better correlation with the geography of the lesion, such programs are not yet widely available.¹²

Reflectance confocal microscopy affords many benefits for patients and providers. It is noninvasive and painless and is capable of imaging in vivo live skin as compared to clinical examination and dermoscopy, which only allow for visualization of the skin’s surface. Reflectance confocal microscopy also is time efficient, as imaging of a single lesion can be completed in 10 to 15 minutes. This technology generates high-resolution images, and RCM diagnosis has consistently demonstrated high sensitivity and specificity when compared to histopathology.¹³ Additionally, RCM imaging can spare biopsy and resultant scarring on cosmetically sensitive areas. Recently, RCM imaging of the skin has been granted Category I Current Procedural Terminology reimbursement codes that allow provider reimbursement and integration of RCM into daily practice¹⁴; however, private insurance coverage in the United States is variable. Limitations of RCM include a maximum depth of 200 to 300 µm, high cost to procure a reflectance confocal microscope, and the need for considerable training and practice to accurately interpret grayscale en face images.¹⁵

There has been extensive research regarding the use of RCM in the evaluation of cutaneous neoplasms and other skin diseases. Numerous features and patterns have been identified and described that correspond with different skin diseases and correspond well with histopathology (Figure 2).^13,16,17 Reflectance confocal microscopy has demonstrated consistently high accuracy in the diagnosis of melanocytic lesions, with a sensitivity of 93% to 100% and a specificity of 75% to 99%.^18-21 Reflectance confocal microscopy is especially useful in the evaluation of clinically or dermoscopically equivocal pigmented lesions due to greater specificity, resulting in a reduction of unnecessary biopsies.^22,23 It also has high accuracy in the diagnosis of keratinocytic neoplasms, with a sensitivity of 82% to 100% and a specificity of 78% to 97% in the diagnosis of BCC,²⁴ and a sensitivity of 74% to 100% and specificity of 78% to 100% in the diagnosis of squamous cell carcinoma (SCC).^25,26 Evaluation of SCC and actinic keratosis (AK) using RCM may be limited by considerable hyperkeratosis and ulceration. In addition, it can be challenging to differentiate AK and SCC on RCM, and considerable expertise is required to accurately grade cytologic and architectural atypia.²⁷ However, RCM has been used to discriminate between in situ and invasive proliferations.²⁸ Reflectance confocal microscopy has wide applications in the diagnosis and management of cutaneous infections^29,30 and inflammatory skin diseases.^29,31-33 Recent RCM research explored the use of RCM to identify biopsy sites,³⁴ delineate presurgical tumor margins,^35,36 and monitor response to noninvasive treatments.^37,38

Optical Coherence Tomography

Optical coherence tomography is an imaging modality that utilizes light backscatter from infrared light to produce grayscale cross-sectional or vertical images and horizontal en face images.³⁹ Optical coherence tomography can visualize structures in the epidermis, dermoepidermal junction, and upper dermis.⁴⁰ It can image boundaries of structures but cannot visualize individual cells.

There are different types of OCT devices available, including frequency-domain OCT (FD-OCT), or conventional OCT, and high-definition OCT (HD-OCT). With FD-OCT, images are captured at a maximum depth of 1 to 2 mm but with limited resolution. High-definition OCT has superior resolution compared to FD-OCT but is restricted to a shallower depth of 750 μm.³⁹ The main advantage of OCT is the ability to noninvasively image live tissue and visualize 2- to 5-times greater depth as compared to RCM. Several OCT devices have obtained US Food and Drug Administration approval; however, OCT has not been widely adopted into clinical practice and is available only in tertiary academic centers. Additionally, OCT imaging in dermatology is rarely reimbursed. Other limitations of OCT include poor resolution of images, high cost to procure an OCT device, and the need for advanced training and experience to accurately interpret images.^40,41

Optical coherence tomography primarily is used to diagnose cutaneous neoplasms. The best evidence of the diagnostic accuracy of OCT is in the setting of BCC, with a recent systematic review reporting a sensitivity of 66% to 96% and a specificity of 75% to 86% for conventional FD-OCT.⁴² The use of FD-OCT results in an increase in specificity without a significant change in sensitivity when compared to dermoscopy in the diagnosis of BCC.⁴³ Melanoma is difficult to diagnose via FD-OCT, as the visualization of architectural features often is limited by poor resolution.⁴⁴ A study of HD-OCT in the diagnosis of melanoma with a limited sample size reported a sensitivity of 74% to 80% and a specificity of 92% to 93%.⁴⁵ Similarly, a study of HD-OCT used in the diagnosis of AK and SCC revealed a sensitivity and specificity of 81.6% and 92.6%, respectively, for AK and 93.8% and 98.9%, respectively, for SCC.⁴⁶

Numerous algorithms and scoring systems have been developed to further explore the utility of OCT in the diagnosis of cutaneous neoplasms.^47,48 Recent research investigated the utility of dynamic OCT, which can evaluate microvasculature in the diagnosis of cutaneous neoplasms (Figure 3)⁴⁹; the combination of OCT with other imaging modalities^50,51; the use of OCT to delineate presurgical margins^52,53; and the role of OCT in the diagnosis and monitoring of inflammatory and infectious skin diseases.^54,55

Final Thoughts

In recent years, there has been a surge of interest in noninvasive techniques for diagnosis and management of skin diseases; however, noninvasive tools exist on a spectrum in dermatology. Dermoscopy provides low-cost imaging of the skin’s surface and has been widely adopted by dermatologists and other providers to aid in clinical diagnosis. Reflectance confocal microscopy provides reimbursable in vivo imaging of live tissue with cellular-level resolution but is limited by depth, cost, and need for advanced training; thus, RCM has only been adopted in some clinical practices. Optical coherence tomography offers in vivo imaging of live tissue with substantial depth but poor resolution, high cost, need for advanced training, and rare reimbursement for providers. Future directions include combination of complementary imaging modalities, increased clinical practice integration, and education and reimbursement for providers.

Traditionally, diagnosis of skin disease relies on clinical inspection, often followed by biopsy and histopathologic examination. In recent years, new noninvasive tools have emerged that can aid in clinical diagnosis and reduce the number of unnecessary benign biopsies. Although there has been a surge in noninvasive diagnostic technologies, many tools are still in research and development phases, with few tools widely adopted and used in regular clinical practice. In this article, we discuss the use of dermoscopy, reflectance confocal microscopy (RCM), and optical coherence tomography (OCT) in the diagnosis and management of skin disease.

Dermoscopy

Dermoscopy, also known as epiluminescence light microscopy and previously known as dermatoscopy, utilizes a ×10 to ×100 microscope objective with a light source to magnify and visualize structures present below the skin’s surface, such as melanin and blood vessels. There are 3 types of dermoscopy: conventional nonpolarized dermoscopy, polarized contact dermoscopy, and nonpolarized contact dermoscopy (Figure 1). Traditional nonpolarized dermoscopy requires a liquid medium and direct contact with the skin, and it relies on light reflection and refraction properties.¹ Cross-polarized light sources allow visualization of deeper structures, either with or without a liquid medium and contact with the skin surface. Although there is overall concurrence among the different types of dermoscopy, subtle differences in the appearance of color, features, and structure are present.¹

Dermoscopy offers many benefits for dermatologists and other providers. It can be used to aid in the diagnosis of cutaneous neoplasms and other skin diseases. Numerous low-cost dermatoscopes currently are commercially available. The handheld, easily transportable nature of dermatoscopes have resulted in widespread practice integration. Approximately 84% of attending dermatologists in US academic settings reported using dermoscopy, and many refer to the dermatoscope as “the dermatologist’s stethoscope.”² In addition, 6% to 15% of other US providers, including family physicians, internal medicine physicians, and plastic surgeons, have reported using dermoscopy in their clinical practices. Limitations of dermoscopy include visualization of the skin surface only and not deeper structures within the tissue, the need for training for adequate interpretation of dermoscopic images, and lack of reimbursement for dermoscopic examination.³

Many dermoscopic structures that correspond well with histopathology have been described. Dermoscopy has a sensitivity of 79% to 96% and specificity of 69% to 99% in the diagnosis of melanoma.⁴ There is variable data on the specificity of dermoscopy in the diagnosis of melanoma, with one meta-analysis finding no statistically significant difference in specificity compared to naked eye examination,⁵ while other studies report increased specificity and subsequent reduction in biopsy of benign lesions.^6,7 Dermoscopy also can aid in the diagnosis of keratinocytic neoplasms, and dermoscopy also results in a sensitivity of 78.6% to 100% and a specificity of 53.8% to 100% in the diagnosis of basal cell carcinoma (BCC).⁸ Limitations of dermoscopy include false-positive diagnoses, commonly seborrheic keratoses and nevi, resulting in unnecessary biopsies, as well as false-negative diagnoses, commonly amelanotic and nevoid melanoma, resulting in delays in skin cancer diagnosis and resultant poor outcomes.⁹ Dermoscopy also is used to aid in the diagnosis of inflammatory and infectious skin diseases, as well as scalp, hair, and nail disorders.¹⁰

Reflectance Confocal Microscopy

Reflectance confocal microscopy utilizes an 830-nm laser to capture horizontal en face images of the skin with high resolution. Different structures of the skin have varying indices of refraction: keratin, melanin, and collagen appear bright white, while other components appear dark, generating black-and-white RCM images.¹¹ Currently, there are 2 reflectance confocal microscopes that are commercially available in the United States. The Vivascope 1500 (Caliber ID) is the traditional model that captures 8×8-mm images, and the Vivascope 3000 (Caliber ID) is a smaller handheld model that captures 0.5×0.5-mm images. The traditional model provides the advantages of higher-resolution images and the ability to capture larger surface areas but is best suited to image flat areas of skin to which a square window can be adhered. The handheld model allows improved contact with the varying topography of skin; does not require an adhesive window; and can be used to image cartilaginous, mucosal, and sensitive surfaces. However, it can be difficult to correlate individual images captured by the handheld RCM with the location relative to the lesion, as it is exquisitely sensitive to motion and also is operator dependent. Although complex algorithms are under development to stitch individual images to provide better correlation with the geography of the lesion, such programs are not yet widely available.¹²

Reflectance confocal microscopy affords many benefits for patients and providers. It is noninvasive and painless and is capable of imaging in vivo live skin as compared to clinical examination and dermoscopy, which only allow for visualization of the skin’s surface. Reflectance confocal microscopy also is time efficient, as imaging of a single lesion can be completed in 10 to 15 minutes. This technology generates high-resolution images, and RCM diagnosis has consistently demonstrated high sensitivity and specificity when compared to histopathology.¹³ Additionally, RCM imaging can spare biopsy and resultant scarring on cosmetically sensitive areas. Recently, RCM imaging of the skin has been granted Category I Current Procedural Terminology reimbursement codes that allow provider reimbursement and integration of RCM into daily practice¹⁴; however, private insurance coverage in the United States is variable. Limitations of RCM include a maximum depth of 200 to 300 µm, high cost to procure a reflectance confocal microscope, and the need for considerable training and practice to accurately interpret grayscale en face images.¹⁵

There has been extensive research regarding the use of RCM in the evaluation of cutaneous neoplasms and other skin diseases. Numerous features and patterns have been identified and described that correspond with different skin diseases and correspond well with histopathology (Figure 2).^13,16,17 Reflectance confocal microscopy has demonstrated consistently high accuracy in the diagnosis of melanocytic lesions, with a sensitivity of 93% to 100% and a specificity of 75% to 99%.^18-21 Reflectance confocal microscopy is especially useful in the evaluation of clinically or dermoscopically equivocal pigmented lesions due to greater specificity, resulting in a reduction of unnecessary biopsies.^22,23 It also has high accuracy in the diagnosis of keratinocytic neoplasms, with a sensitivity of 82% to 100% and a specificity of 78% to 97% in the diagnosis of BCC,²⁴ and a sensitivity of 74% to 100% and specificity of 78% to 100% in the diagnosis of squamous cell carcinoma (SCC).^25,26 Evaluation of SCC and actinic keratosis (AK) using RCM may be limited by considerable hyperkeratosis and ulceration. In addition, it can be challenging to differentiate AK and SCC on RCM, and considerable expertise is required to accurately grade cytologic and architectural atypia.²⁷ However, RCM has been used to discriminate between in situ and invasive proliferations.²⁸ Reflectance confocal microscopy has wide applications in the diagnosis and management of cutaneous infections^29,30 and inflammatory skin diseases.^29,31-33 Recent RCM research explored the use of RCM to identify biopsy sites,³⁴ delineate presurgical tumor margins,^35,36 and monitor response to noninvasive treatments.^37,38

Optical Coherence Tomography

Optical coherence tomography is an imaging modality that utilizes light backscatter from infrared light to produce grayscale cross-sectional or vertical images and horizontal en face images.³⁹ Optical coherence tomography can visualize structures in the epidermis, dermoepidermal junction, and upper dermis.⁴⁰ It can image boundaries of structures but cannot visualize individual cells.

There are different types of OCT devices available, including frequency-domain OCT (FD-OCT), or conventional OCT, and high-definition OCT (HD-OCT). With FD-OCT, images are captured at a maximum depth of 1 to 2 mm but with limited resolution. High-definition OCT has superior resolution compared to FD-OCT but is restricted to a shallower depth of 750 μm.³⁹ The main advantage of OCT is the ability to noninvasively image live tissue and visualize 2- to 5-times greater depth as compared to RCM. Several OCT devices have obtained US Food and Drug Administration approval; however, OCT has not been widely adopted into clinical practice and is available only in tertiary academic centers. Additionally, OCT imaging in dermatology is rarely reimbursed. Other limitations of OCT include poor resolution of images, high cost to procure an OCT device, and the need for advanced training and experience to accurately interpret images.^40,41

Optical coherence tomography primarily is used to diagnose cutaneous neoplasms. The best evidence of the diagnostic accuracy of OCT is in the setting of BCC, with a recent systematic review reporting a sensitivity of 66% to 96% and a specificity of 75% to 86% for conventional FD-OCT.⁴² The use of FD-OCT results in an increase in specificity without a significant change in sensitivity when compared to dermoscopy in the diagnosis of BCC.⁴³ Melanoma is difficult to diagnose via FD-OCT, as the visualization of architectural features often is limited by poor resolution.⁴⁴ A study of HD-OCT in the diagnosis of melanoma with a limited sample size reported a sensitivity of 74% to 80% and a specificity of 92% to 93%.⁴⁵ Similarly, a study of HD-OCT used in the diagnosis of AK and SCC revealed a sensitivity and specificity of 81.6% and 92.6%, respectively, for AK and 93.8% and 98.9%, respectively, for SCC.⁴⁶

Numerous algorithms and scoring systems have been developed to further explore the utility of OCT in the diagnosis of cutaneous neoplasms.^47,48 Recent research investigated the utility of dynamic OCT, which can evaluate microvasculature in the diagnosis of cutaneous neoplasms (Figure 3)⁴⁹; the combination of OCT with other imaging modalities^50,51; the use of OCT to delineate presurgical margins^52,53; and the role of OCT in the diagnosis and monitoring of inflammatory and infectious skin diseases.^54,55

Final Thoughts

In recent years, there has been a surge of interest in noninvasive techniques for diagnosis and management of skin diseases; however, noninvasive tools exist on a spectrum in dermatology. Dermoscopy provides low-cost imaging of the skin’s surface and has been widely adopted by dermatologists and other providers to aid in clinical diagnosis. Reflectance confocal microscopy provides reimbursable in vivo imaging of live tissue with cellular-level resolution but is limited by depth, cost, and need for advanced training; thus, RCM has only been adopted in some clinical practices. Optical coherence tomography offers in vivo imaging of live tissue with substantial depth but poor resolution, high cost, need for advanced training, and rare reimbursement for providers. Future directions include combination of complementary imaging modalities, increased clinical practice integration, and education and reimbursement for providers.

Traditionally, diagnosis of skin disease relies on clinical inspection, often followed by biopsy and histopathologic examination. In recent years, new noninvasive tools have emerged that can aid in clinical diagnosis and reduce the number of unnecessary benign biopsies. Although there has been a surge in noninvasive diagnostic technologies, many tools are still in research and development phases, with few tools widely adopted and used in regular clinical practice. In this article, we discuss the use of dermoscopy, reflectance confocal microscopy (RCM), and optical coherence tomography (OCT) in the diagnosis and management of skin disease.

Dermoscopy

Dermoscopy, also known as epiluminescence light microscopy and previously known as dermatoscopy, utilizes a ×10 to ×100 microscope objective with a light source to magnify and visualize structures present below the skin’s surface, such as melanin and blood vessels. There are 3 types of dermoscopy: conventional nonpolarized dermoscopy, polarized contact dermoscopy, and nonpolarized contact dermoscopy (Figure 1). Traditional nonpolarized dermoscopy requires a liquid medium and direct contact with the skin, and it relies on light reflection and refraction properties.¹ Cross-polarized light sources allow visualization of deeper structures, either with or without a liquid medium and contact with the skin surface. Although there is overall concurrence among the different types of dermoscopy, subtle differences in the appearance of color, features, and structure are present.¹

Dermoscopy offers many benefits for dermatologists and other providers. It can be used to aid in the diagnosis of cutaneous neoplasms and other skin diseases. Numerous low-cost dermatoscopes currently are commercially available. The handheld, easily transportable nature of dermatoscopes have resulted in widespread practice integration. Approximately 84% of attending dermatologists in US academic settings reported using dermoscopy, and many refer to the dermatoscope as “the dermatologist’s stethoscope.”² In addition, 6% to 15% of other US providers, including family physicians, internal medicine physicians, and plastic surgeons, have reported using dermoscopy in their clinical practices. Limitations of dermoscopy include visualization of the skin surface only and not deeper structures within the tissue, the need for training for adequate interpretation of dermoscopic images, and lack of reimbursement for dermoscopic examination.³

Many dermoscopic structures that correspond well with histopathology have been described. Dermoscopy has a sensitivity of 79% to 96% and specificity of 69% to 99% in the diagnosis of melanoma.⁴ There is variable data on the specificity of dermoscopy in the diagnosis of melanoma, with one meta-analysis finding no statistically significant difference in specificity compared to naked eye examination,⁵ while other studies report increased specificity and subsequent reduction in biopsy of benign lesions.^6,7 Dermoscopy also can aid in the diagnosis of keratinocytic neoplasms, and dermoscopy also results in a sensitivity of 78.6% to 100% and a specificity of 53.8% to 100% in the diagnosis of basal cell carcinoma (BCC).⁸ Limitations of dermoscopy include false-positive diagnoses, commonly seborrheic keratoses and nevi, resulting in unnecessary biopsies, as well as false-negative diagnoses, commonly amelanotic and nevoid melanoma, resulting in delays in skin cancer diagnosis and resultant poor outcomes.⁹ Dermoscopy also is used to aid in the diagnosis of inflammatory and infectious skin diseases, as well as scalp, hair, and nail disorders.¹⁰

Reflectance Confocal Microscopy

Reflectance confocal microscopy utilizes an 830-nm laser to capture horizontal en face images of the skin with high resolution. Different structures of the skin have varying indices of refraction: keratin, melanin, and collagen appear bright white, while other components appear dark, generating black-and-white RCM images.¹¹ Currently, there are 2 reflectance confocal microscopes that are commercially available in the United States. The Vivascope 1500 (Caliber ID) is the traditional model that captures 8×8-mm images, and the Vivascope 3000 (Caliber ID) is a smaller handheld model that captures 0.5×0.5-mm images. The traditional model provides the advantages of higher-resolution images and the ability to capture larger surface areas but is best suited to image flat areas of skin to which a square window can be adhered. The handheld model allows improved contact with the varying topography of skin; does not require an adhesive window; and can be used to image cartilaginous, mucosal, and sensitive surfaces. However, it can be difficult to correlate individual images captured by the handheld RCM with the location relative to the lesion, as it is exquisitely sensitive to motion and also is operator dependent. Although complex algorithms are under development to stitch individual images to provide better correlation with the geography of the lesion, such programs are not yet widely available.¹²

Reflectance confocal microscopy affords many benefits for patients and providers. It is noninvasive and painless and is capable of imaging in vivo live skin as compared to clinical examination and dermoscopy, which only allow for visualization of the skin’s surface. Reflectance confocal microscopy also is time efficient, as imaging of a single lesion can be completed in 10 to 15 minutes. This technology generates high-resolution images, and RCM diagnosis has consistently demonstrated high sensitivity and specificity when compared to histopathology.¹³ Additionally, RCM imaging can spare biopsy and resultant scarring on cosmetically sensitive areas. Recently, RCM imaging of the skin has been granted Category I Current Procedural Terminology reimbursement codes that allow provider reimbursement and integration of RCM into daily practice¹⁴; however, private insurance coverage in the United States is variable. Limitations of RCM include a maximum depth of 200 to 300 µm, high cost to procure a reflectance confocal microscope, and the need for considerable training and practice to accurately interpret grayscale en face images.¹⁵

There has been extensive research regarding the use of RCM in the evaluation of cutaneous neoplasms and other skin diseases. Numerous features and patterns have been identified and described that correspond with different skin diseases and correspond well with histopathology (Figure 2).^13,16,17 Reflectance confocal microscopy has demonstrated consistently high accuracy in the diagnosis of melanocytic lesions, with a sensitivity of 93% to 100% and a specificity of 75% to 99%.^18-21 Reflectance confocal microscopy is especially useful in the evaluation of clinically or dermoscopically equivocal pigmented lesions due to greater specificity, resulting in a reduction of unnecessary biopsies.^22,23 It also has high accuracy in the diagnosis of keratinocytic neoplasms, with a sensitivity of 82% to 100% and a specificity of 78% to 97% in the diagnosis of BCC,²⁴ and a sensitivity of 74% to 100% and specificity of 78% to 100% in the diagnosis of squamous cell carcinoma (SCC).^25,26 Evaluation of SCC and actinic keratosis (AK) using RCM may be limited by considerable hyperkeratosis and ulceration. In addition, it can be challenging to differentiate AK and SCC on RCM, and considerable expertise is required to accurately grade cytologic and architectural atypia.²⁷ However, RCM has been used to discriminate between in situ and invasive proliferations.²⁸ Reflectance confocal microscopy has wide applications in the diagnosis and management of cutaneous infections^29,30 and inflammatory skin diseases.^29,31-33 Recent RCM research explored the use of RCM to identify biopsy sites,³⁴ delineate presurgical tumor margins,^35,36 and monitor response to noninvasive treatments.^37,38

Optical Coherence Tomography

Optical coherence tomography is an imaging modality that utilizes light backscatter from infrared light to produce grayscale cross-sectional or vertical images and horizontal en face images.³⁹ Optical coherence tomography can visualize structures in the epidermis, dermoepidermal junction, and upper dermis.⁴⁰ It can image boundaries of structures but cannot visualize individual cells.

There are different types of OCT devices available, including frequency-domain OCT (FD-OCT), or conventional OCT, and high-definition OCT (HD-OCT). With FD-OCT, images are captured at a maximum depth of 1 to 2 mm but with limited resolution. High-definition OCT has superior resolution compared to FD-OCT but is restricted to a shallower depth of 750 μm.³⁹ The main advantage of OCT is the ability to noninvasively image live tissue and visualize 2- to 5-times greater depth as compared to RCM. Several OCT devices have obtained US Food and Drug Administration approval; however, OCT has not been widely adopted into clinical practice and is available only in tertiary academic centers. Additionally, OCT imaging in dermatology is rarely reimbursed. Other limitations of OCT include poor resolution of images, high cost to procure an OCT device, and the need for advanced training and experience to accurately interpret images.^40,41

Optical coherence tomography primarily is used to diagnose cutaneous neoplasms. The best evidence of the diagnostic accuracy of OCT is in the setting of BCC, with a recent systematic review reporting a sensitivity of 66% to 96% and a specificity of 75% to 86% for conventional FD-OCT.⁴² The use of FD-OCT results in an increase in specificity without a significant change in sensitivity when compared to dermoscopy in the diagnosis of BCC.⁴³ Melanoma is difficult to diagnose via FD-OCT, as the visualization of architectural features often is limited by poor resolution.⁴⁴ A study of HD-OCT in the diagnosis of melanoma with a limited sample size reported a sensitivity of 74% to 80% and a specificity of 92% to 93%.⁴⁵ Similarly, a study of HD-OCT used in the diagnosis of AK and SCC revealed a sensitivity and specificity of 81.6% and 92.6%, respectively, for AK and 93.8% and 98.9%, respectively, for SCC.⁴⁶

Numerous algorithms and scoring systems have been developed to further explore the utility of OCT in the diagnosis of cutaneous neoplasms.^47,48 Recent research investigated the utility of dynamic OCT, which can evaluate microvasculature in the diagnosis of cutaneous neoplasms (Figure 3)⁴⁹; the combination of OCT with other imaging modalities^50,51; the use of OCT to delineate presurgical margins^52,53; and the role of OCT in the diagnosis and monitoring of inflammatory and infectious skin diseases.^54,55

Final Thoughts

In recent years, there has been a surge of interest in noninvasive techniques for diagnosis and management of skin diseases; however, noninvasive tools exist on a spectrum in dermatology. Dermoscopy provides low-cost imaging of the skin’s surface and has been widely adopted by dermatologists and other providers to aid in clinical diagnosis. Reflectance confocal microscopy provides reimbursable in vivo imaging of live tissue with cellular-level resolution but is limited by depth, cost, and need for advanced training; thus, RCM has only been adopted in some clinical practices. Optical coherence tomography offers in vivo imaging of live tissue with substantial depth but poor resolution, high cost, need for advanced training, and rare reimbursement for providers. Future directions include combination of complementary imaging modalities, increased clinical practice integration, and education and reimbursement for providers.

The advent of electronic medical records (EMRs) is arguably the most important technological revolution in modern medicine. The transition from paper documentation to EMRs has improved organization of medical records, consolidating all physician notes, orders, consultations, laboratory test results, and radiologic studies into a single accessible location.¹ However, this revolution has led to mixed consequences for patients, especially in the outpatient setting. The use of EMRs can facilitate questions, clarification, and discussion between patients and health care providers, prompted by the sections of the EMR. Unfortunately, patients too often encounter pressed-for-time, documentation-focused providers who may not even look up from the computer. Provider behaviors such as making eye contact, stopping typing during discussion of sensitive topics, and allowing patients to view the computer screen and using it as an educational tool are important for patients to have a positive care experience.² We envision further integration of current and future technology to overcome the challenges of outpatient care. We use a hypothetical patient encounter to illustrate what the future may hold.

Hypothetical Patient Encounter

An established patient, Ms. PS, comes to the dermatology clinic for a follow-up appointment and walks into an examination room (Figure). Prior to entering the room, the provider, Dr. FT, reviews Ms. PS’s history via a dermatology-specific EMR and reads that Ms. PS has a 1.5-year history of psoriasis and is considering other therapeutic options.

Upon entering the room, Dr. FT tells Ms. PS that the visit is being recorded and transcribed. A large interactive screen is a key component of the examination room. A remote medical assistant is virtually present via video to transcribe and document the patient-provider interaction. There is potential for artificial intelligence to replace the remote medical assistant in the future. Wearable technology, including a smartwatch and Bluetooth headphones, allow the provider to record audio of the visit as well as through microphones on the interactive screen.

As the interaction begins, Ms. PS reports that her psoriasis is poorly controlled with her current regimen of topical steroids. Dr. FT inquires about Ms. PS’s current symptoms and psychosocial well-being. Dr. FT then performs a skin examination and is easily able to evaluate her skin vs prior visits, as clinical images from prior visits are automatically displayed on the interactive screen. Dr. FT also closely examines Ms. PS’s nails and conducts a joint examination, reminded by a notification on his wearable technology. After capturing clinical images of Ms. PS’s skin and nails with a secure EMR-connected tablet, Dr. FT briefly steps out of the room to allow Ms. PS to get dressed and feel more comfortable in the discussion to follow.

Once he reenters the examination room, Dr. FT initiates a discussion on next steps. Ms. PS’s pathology report and clinical images are displayed on the interactive screen, along with her most recent laboratory test results, which were completed prior to the visit in anticipation of changing therapies. Dr. FT presents Ms. PS with several evidence-based therapeutic options for psoriasis, and she expresses interest in methotrexate. Following the discussion, the remote medical assistant displays information about methotrexate on the interactive screen, including evidence for treatment of psoriasis, contraindications, laboratory monitoring requirements, and possible adverse effects for both the patient and provider to review together. Dr. FT reviews the laboratory test results displayed on the screen, specifically her transaminase levels, and confirms that methotrexate is an appropriate therapeutic option. After a full discussion of risks and benefits, Ms. PS chooses to initiate methotrexate treatment. Reminded by a notification on his wearable technology, Dr. FT follows evidence-based dosing guidelines and sends the prescription electronically to Ms. PS’s pharmacy, which concludes Ms. PS’s visit.

Analysis of the Patient Encounter

In this interaction, Dr. FT was able to fully engage with the patient, unencumbered by the demands of documentation. There were only a few instances when the provider looked at or touched the interactive screen. Furthermore, joint decision-making was optimized by allowing both the patient and provider to review diagnostic test results and current evidence-based therapeutic guidelines together through the interactive screen. Ms. PS goes home feeling satisfied that she received her provider’s complete attention and that they selected a therapeutic option supported by evidence. After the visit, the remote medial assistant’s transcript populates a patient note template, which Dr. FT reviews and amends to create the final note. Reducing the time required to write patient notes increases the speed at which Dr. FT can complete patient encounters and may improve clinic flow and productivity. In addition, a patient summary is generated from Dr. FT’s final note, with an emphasis on patient instructions, and is sent to Ms. PS.

Final Thoughts

Our proposed integration of currently available and future technology can help minimize documentation burdens on providers and improve patient-provider communication in the age of the EMR, thus optimizing patient satisfaction and outcomes.

The advent of electronic medical records (EMRs) is arguably the most important technological revolution in modern medicine. The transition from paper documentation to EMRs has improved organization of medical records, consolidating all physician notes, orders, consultations, laboratory test results, and radiologic studies into a single accessible location.¹ However, this revolution has led to mixed consequences for patients, especially in the outpatient setting. The use of EMRs can facilitate questions, clarification, and discussion between patients and health care providers, prompted by the sections of the EMR. Unfortunately, patients too often encounter pressed-for-time, documentation-focused providers who may not even look up from the computer. Provider behaviors such as making eye contact, stopping typing during discussion of sensitive topics, and allowing patients to view the computer screen and using it as an educational tool are important for patients to have a positive care experience.² We envision further integration of current and future technology to overcome the challenges of outpatient care. We use a hypothetical patient encounter to illustrate what the future may hold.

Hypothetical Patient Encounter

An established patient, Ms. PS, comes to the dermatology clinic for a follow-up appointment and walks into an examination room (Figure). Prior to entering the room, the provider, Dr. FT, reviews Ms. PS’s history via a dermatology-specific EMR and reads that Ms. PS has a 1.5-year history of psoriasis and is considering other therapeutic options.

Upon entering the room, Dr. FT tells Ms. PS that the visit is being recorded and transcribed. A large interactive screen is a key component of the examination room. A remote medical assistant is virtually present via video to transcribe and document the patient-provider interaction. There is potential for artificial intelligence to replace the remote medical assistant in the future. Wearable technology, including a smartwatch and Bluetooth headphones, allow the provider to record audio of the visit as well as through microphones on the interactive screen.

As the interaction begins, Ms. PS reports that her psoriasis is poorly controlled with her current regimen of topical steroids. Dr. FT inquires about Ms. PS’s current symptoms and psychosocial well-being. Dr. FT then performs a skin examination and is easily able to evaluate her skin vs prior visits, as clinical images from prior visits are automatically displayed on the interactive screen. Dr. FT also closely examines Ms. PS’s nails and conducts a joint examination, reminded by a notification on his wearable technology. After capturing clinical images of Ms. PS’s skin and nails with a secure EMR-connected tablet, Dr. FT briefly steps out of the room to allow Ms. PS to get dressed and feel more comfortable in the discussion to follow.

Once he reenters the examination room, Dr. FT initiates a discussion on next steps. Ms. PS’s pathology report and clinical images are displayed on the interactive screen, along with her most recent laboratory test results, which were completed prior to the visit in anticipation of changing therapies. Dr. FT presents Ms. PS with several evidence-based therapeutic options for psoriasis, and she expresses interest in methotrexate. Following the discussion, the remote medical assistant displays information about methotrexate on the interactive screen, including evidence for treatment of psoriasis, contraindications, laboratory monitoring requirements, and possible adverse effects for both the patient and provider to review together. Dr. FT reviews the laboratory test results displayed on the screen, specifically her transaminase levels, and confirms that methotrexate is an appropriate therapeutic option. After a full discussion of risks and benefits, Ms. PS chooses to initiate methotrexate treatment. Reminded by a notification on his wearable technology, Dr. FT follows evidence-based dosing guidelines and sends the prescription electronically to Ms. PS’s pharmacy, which concludes Ms. PS’s visit.

Analysis of the Patient Encounter

In this interaction, Dr. FT was able to fully engage with the patient, unencumbered by the demands of documentation. There were only a few instances when the provider looked at or touched the interactive screen. Furthermore, joint decision-making was optimized by allowing both the patient and provider to review diagnostic test results and current evidence-based therapeutic guidelines together through the interactive screen. Ms. PS goes home feeling satisfied that she received her provider’s complete attention and that they selected a therapeutic option supported by evidence. After the visit, the remote medial assistant’s transcript populates a patient note template, which Dr. FT reviews and amends to create the final note. Reducing the time required to write patient notes increases the speed at which Dr. FT can complete patient encounters and may improve clinic flow and productivity. In addition, a patient summary is generated from Dr. FT’s final note, with an emphasis on patient instructions, and is sent to Ms. PS.

Final Thoughts

Our proposed integration of currently available and future technology can help minimize documentation burdens on providers and improve patient-provider communication in the age of the EMR, thus optimizing patient satisfaction and outcomes.

The advent of electronic medical records (EMRs) is arguably the most important technological revolution in modern medicine. The transition from paper documentation to EMRs has improved organization of medical records, consolidating all physician notes, orders, consultations, laboratory test results, and radiologic studies into a single accessible location.¹ However, this revolution has led to mixed consequences for patients, especially in the outpatient setting. The use of EMRs can facilitate questions, clarification, and discussion between patients and health care providers, prompted by the sections of the EMR. Unfortunately, patients too often encounter pressed-for-time, documentation-focused providers who may not even look up from the computer. Provider behaviors such as making eye contact, stopping typing during discussion of sensitive topics, and allowing patients to view the computer screen and using it as an educational tool are important for patients to have a positive care experience.² We envision further integration of current and future technology to overcome the challenges of outpatient care. We use a hypothetical patient encounter to illustrate what the future may hold.

Hypothetical Patient Encounter

An established patient, Ms. PS, comes to the dermatology clinic for a follow-up appointment and walks into an examination room (Figure). Prior to entering the room, the provider, Dr. FT, reviews Ms. PS’s history via a dermatology-specific EMR and reads that Ms. PS has a 1.5-year history of psoriasis and is considering other therapeutic options.

Upon entering the room, Dr. FT tells Ms. PS that the visit is being recorded and transcribed. A large interactive screen is a key component of the examination room. A remote medical assistant is virtually present via video to transcribe and document the patient-provider interaction. There is potential for artificial intelligence to replace the remote medical assistant in the future. Wearable technology, including a smartwatch and Bluetooth headphones, allow the provider to record audio of the visit as well as through microphones on the interactive screen.

As the interaction begins, Ms. PS reports that her psoriasis is poorly controlled with her current regimen of topical steroids. Dr. FT inquires about Ms. PS’s current symptoms and psychosocial well-being. Dr. FT then performs a skin examination and is easily able to evaluate her skin vs prior visits, as clinical images from prior visits are automatically displayed on the interactive screen. Dr. FT also closely examines Ms. PS’s nails and conducts a joint examination, reminded by a notification on his wearable technology. After capturing clinical images of Ms. PS’s skin and nails with a secure EMR-connected tablet, Dr. FT briefly steps out of the room to allow Ms. PS to get dressed and feel more comfortable in the discussion to follow.

Once he reenters the examination room, Dr. FT initiates a discussion on next steps. Ms. PS’s pathology report and clinical images are displayed on the interactive screen, along with her most recent laboratory test results, which were completed prior to the visit in anticipation of changing therapies. Dr. FT presents Ms. PS with several evidence-based therapeutic options for psoriasis, and she expresses interest in methotrexate. Following the discussion, the remote medical assistant displays information about methotrexate on the interactive screen, including evidence for treatment of psoriasis, contraindications, laboratory monitoring requirements, and possible adverse effects for both the patient and provider to review together. Dr. FT reviews the laboratory test results displayed on the screen, specifically her transaminase levels, and confirms that methotrexate is an appropriate therapeutic option. After a full discussion of risks and benefits, Ms. PS chooses to initiate methotrexate treatment. Reminded by a notification on his wearable technology, Dr. FT follows evidence-based dosing guidelines and sends the prescription electronically to Ms. PS’s pharmacy, which concludes Ms. PS’s visit.

Analysis of the Patient Encounter

In this interaction, Dr. FT was able to fully engage with the patient, unencumbered by the demands of documentation. There were only a few instances when the provider looked at or touched the interactive screen. Furthermore, joint decision-making was optimized by allowing both the patient and provider to review diagnostic test results and current evidence-based therapeutic guidelines together through the interactive screen. Ms. PS goes home feeling satisfied that she received her provider’s complete attention and that they selected a therapeutic option supported by evidence. After the visit, the remote medial assistant’s transcript populates a patient note template, which Dr. FT reviews and amends to create the final note. Reducing the time required to write patient notes increases the speed at which Dr. FT can complete patient encounters and may improve clinic flow and productivity. In addition, a patient summary is generated from Dr. FT’s final note, with an emphasis on patient instructions, and is sent to Ms. PS.

Final Thoughts

Our proposed integration of currently available and future technology can help minimize documentation burdens on providers and improve patient-provider communication in the age of the EMR, thus optimizing patient satisfaction and outcomes.

As the US population continues to grow and patients become more aware of their health needs, payers are beginning to recognize the benefits of more efficient and cost-effective health care. With the implementation of the new Medicare Physician Fee Schedule on January 1, 2019, some old billing codes were revalued while others were replaced entirely with new codes.¹ The restructuring of the standard biopsy codes now takes the complexity of different sampling techniques into consideration. Furthermore, Current Procedural Terminology (CPT) Category III tracking codes for some imaging devices (eg, optical coherence tomography) added in 2017 require more data before obtaining a Category I reimbursable code, while codes for other imaging devices such as reflectance confocal microscopy (RCM) remain relatively the same.^2-4 Notably, the majority of the new 2019 telemedicine codes are applicable to dermatology.^2,3 In this article, we discuss the new CPT codes for reporting diagnostic procedures, including biopsy, noninvasive imaging, and telemedicine services. We also provide a summary of the national average reimbursement rates for these procedures. 

Background on Reimbursement 

To better understand how reimbursement works, it is important to know that all billing codes are provided a relative value unit (RVU), a number representing the value of the work involved and cost of providing a service relative to other services.⁵ The total RVU consists of the work RVU (wRVU), practice expense RVU (peRVU), and malpractice expense RVU (mRVU). The wRVU represents the time, effort, and complexity involved in performing the service. The peRVU reflects the direct cost of supplies, personnel, and durable equipment involved in providing the service, excluding typical office overhead costs such as rent, utilities, and administrative staff. The mRVU is to cover the cost of malpractice insurance.⁵ The peRVU can be further specified as facility versus nonfacility services depending on where the service is performed.⁶ A facility peRVU is for services completed in a facility such as a hospital, outpatient hospital setting, or nursing home. The facility provides some of the involved supplies, personnel, and equipment for which they can recapture costs by separate reporting, resulting in a lower total RVU for the provider charges compared with nonfacility locations where the physician must provide these items.⁶ Many physicians may not be aware of how critical their role is in determining their own reimbursement rates by understanding RVUs and properly filling out Relative Value Scale Update Committee (RUC) surveys. If surveys sent to practitioners are accurately completed, RVUs have the potential to be fairly valued; however, if respondents are unaware of all of the components that are inherent to a procedure, they may end up minimizing the effort or time involved, which would skew the results and hurt those who perform the procedure. Rather than inputting appropriate preoperative and postoperative service times, many respondents often put 0s and 1s throughout the survey, which misrepresents the amount of time involved for a procedure. For example, inputting a preoperative time as 0 or 1 minute may severely underestimate the work involved for a procedure if the true preoperative time is 5 minutes. Such survey responses affect whether or not RVUs are valued appropriately. 

The billing codes and their RVUs as well as Medicare payment values in your area can be found on the Centers for Medicare & Medicaid Services website.^2,3 Table 1 provides a comparison of the old and new biopsy codes, and Table 2 shows the new RCM codes. 

Biopsy Codes 

Prior to 2019, biopsies were reimbursed using CPT code 11100 for the initial biopsy and 11101 for each additional biopsy.² Called up for refinement in the RUC process, initial data from the Physician Practice Expense Information Survey pointed to the likelihood of different sampling techniques having different amounts of work being supplied by different techniques.¹ Imaging modalities such as dermoscopy or RCM could help minimize the need for surgical biopsies. Dermoscopy, which has been proven to allow for more efficient and accurate diagnoses in dermatology, is reimbursed in Europe but not in the United States.^7-9 In 2016, CPT codes 96931 through 96936 were created for RCM and are covered by most insurances.¹⁰ Optical coherence tomography, another noninvasive imaging technology, currently is not reimbursed but did receive Category III codes (0470T-0471T), also known as a tracking codes, in 2017.⁴ Category III codes are used for emerging technologies that have future potential but do not have enough US-based evidence to support receiving Category I CPT codes. The use of Category III codes allows for data collection on emerging technologies and services, with the potential to convert the Category III codes to Category I codes once certain criteria are met.¹¹ 

Beginning in 2019, the standard biopsy codes 11100 and 11101 were replaced with 6 new codes to represent primary (11102, 11104, 11106) and add-on biopsies (11103, 11105, 11107) based on the sampling technique utilized and the thickness of the sample (Table 1). Previously, the biopsy codes did not reflect the complexity of the different biopsy techniques, whereas the new codes provide differentiation of the method of removal (ie, tangential, punch, incisional).^2,3 The base code is dependent on whichever biopsy performed has the highest complexity, with incisional biopsy--a partial excision--being considered the most complex.³ Punch biopsy is considered the next level of complexity, followed by tangential biopsy. Each of the 6 new biopsy codes also received a new wRVU, which determines reimbursement under Medicare and most other insurers when combined with direct peRVU and mRVU. Additional biopsies, reported using the add-on codes, are reimbursed at a lower level than the base codes because of removal of duplicate inputs for preservice and postservice care.³  

Telehealth Codes 

Telemedicine services offer another form of imaging that providers can use to communicate remotely with patients through a live interactive video stream (with audio), a store-and-forward system with photographs or videos shared asynchronously, or remote patient monitoring.¹² Although live video streaming uses a webcam, store-and-forward services involve sending photographs or videos electronically for later evaluation.^12,13 Remote patient monitoring allows the collection of health-related data and transmission to a physician without the need for an office visit.¹³ Most states require physicians to have a license in the state in which the patient is located at the time of the encounter. Given the difficulty of applying for licensure in multiple states, several states started creating their own special licenses to allow out-of-state providers to offer services through telemedicine.¹⁴ The Federation of State Medical Boards then created the Interstate Medical Licensure Compact (IMLC) for an expedited process to apply for medical licensure in other states. The IMLC was formed to increase access to health care in underserved or rural areas including but not limited to the use of telemedicine.¹⁵ To qualify for IMLC, a physician must have a medical license in a state registered with the IMLC (ie, state of principal license) and have at least one of the following in their state of principal license: primary residence, 25% of their medical practice, a current employer, or US federal income taxes filed.¹⁵ The remaining states that do not have a licensing process for telemedicine allow practice in contiguous states or may provide temporary licenses dependent on the situation.¹⁴ 

Since 2017, billing codes for telemedicine have been the same as those used for in-person evaluation and management services with modifiers -95 or GQ added to the end of the code. Modifier -95 has been used for real-time telemedicine services, while modifier GQ has been used for store-and-forward services.¹⁶ For example, the code 99201, which is used to bill for new patients at outpatient visits, would become 99201-95 if performed using a live audio and video feed or 99201-GQ if information was sent electronically for later analysis. To receive reimbursement from Medicare, modifier -95 requires real-time communication using both audio and video; however, modifier GQ is only reimbursable in federal telemedicine demonstration programs in Alaska or Hawaii.¹² Note that reimbursement is up to the discretion of private providers, and even Medicare reimbursement can vary from state to state. 

In 2019, new Healthcare Common Procedure Coding System telemedicine codes were introduced to include virtual check-ins (G2012) and evaluation of patient-transmitted images and videos (G2010). G2010 is the first store-and-forward code that has the potential to be reimbursed outside of Alaska or Hawaii.^3,12 G2012 allows providers to monitor the patients' well-being outside of the office setting, a cost-effective alternative if patients do not require a full visit. More detailed descriptions of the new codes can be found in Table 3.¹ 

Final Thoughts 

As insurance providers continue to better monitor health care costs, it is of utmost importance that physicians become more involved in accurately assessing their services and procedures, given that the changes in RVUs mirror the Centers for Medicare & Medicaid Services' utilization of the RUC's interpretation of our survey responses.¹ The current billing codes attempt to better represent the work involved for each service, one example being the modification to more specific biopsy codes in 2019.  

With the growth of technology, CPT and Healthcare Common Procedure Coding System codes also reflect a push toward more efficient health care delivery and broader coverage for provider services, as demonstrated by the introduction of new telemedicine codes as well as recent additions of noninvasive imaging codes. Although technology makes health care more cost-effective for patients, clinicians can still maintain their overall reimbursements by efficiently seeing an increasing number of patients; for example, a patient diagnosed noninvasively using RCM can then receive same-day care, which impacts patients' quality of life by minimizing travel time, number of office visits, and time taken off from work, while allowing providers to manage a higher patient volume more productively. The new CPT codes discussed here reflect the growth of medical technology potential, which increases our diagnostic capability, making it even more critical for physicians to engage with these developments. 
 

As the US population continues to grow and patients become more aware of their health needs, payers are beginning to recognize the benefits of more efficient and cost-effective health care. With the implementation of the new Medicare Physician Fee Schedule on January 1, 2019, some old billing codes were revalued while others were replaced entirely with new codes.¹ The restructuring of the standard biopsy codes now takes the complexity of different sampling techniques into consideration. Furthermore, Current Procedural Terminology (CPT) Category III tracking codes for some imaging devices (eg, optical coherence tomography) added in 2017 require more data before obtaining a Category I reimbursable code, while codes for other imaging devices such as reflectance confocal microscopy (RCM) remain relatively the same.^2-4 Notably, the majority of the new 2019 telemedicine codes are applicable to dermatology.^2,3 In this article, we discuss the new CPT codes for reporting diagnostic procedures, including biopsy, noninvasive imaging, and telemedicine services. We also provide a summary of the national average reimbursement rates for these procedures. 

Background on Reimbursement 

To better understand how reimbursement works, it is important to know that all billing codes are provided a relative value unit (RVU), a number representing the value of the work involved and cost of providing a service relative to other services.⁵ The total RVU consists of the work RVU (wRVU), practice expense RVU (peRVU), and malpractice expense RVU (mRVU). The wRVU represents the time, effort, and complexity involved in performing the service. The peRVU reflects the direct cost of supplies, personnel, and durable equipment involved in providing the service, excluding typical office overhead costs such as rent, utilities, and administrative staff. The mRVU is to cover the cost of malpractice insurance.⁵ The peRVU can be further specified as facility versus nonfacility services depending on where the service is performed.⁶ A facility peRVU is for services completed in a facility such as a hospital, outpatient hospital setting, or nursing home. The facility provides some of the involved supplies, personnel, and equipment for which they can recapture costs by separate reporting, resulting in a lower total RVU for the provider charges compared with nonfacility locations where the physician must provide these items.⁶ Many physicians may not be aware of how critical their role is in determining their own reimbursement rates by understanding RVUs and properly filling out Relative Value Scale Update Committee (RUC) surveys. If surveys sent to practitioners are accurately completed, RVUs have the potential to be fairly valued; however, if respondents are unaware of all of the components that are inherent to a procedure, they may end up minimizing the effort or time involved, which would skew the results and hurt those who perform the procedure. Rather than inputting appropriate preoperative and postoperative service times, many respondents often put 0s and 1s throughout the survey, which misrepresents the amount of time involved for a procedure. For example, inputting a preoperative time as 0 or 1 minute may severely underestimate the work involved for a procedure if the true preoperative time is 5 minutes. Such survey responses affect whether or not RVUs are valued appropriately. 

The billing codes and their RVUs as well as Medicare payment values in your area can be found on the Centers for Medicare & Medicaid Services website.^2,3 Table 1 provides a comparison of the old and new biopsy codes, and Table 2 shows the new RCM codes. 

Biopsy Codes 

Prior to 2019, biopsies were reimbursed using CPT code 11100 for the initial biopsy and 11101 for each additional biopsy.² Called up for refinement in the RUC process, initial data from the Physician Practice Expense Information Survey pointed to the likelihood of different sampling techniques having different amounts of work being supplied by different techniques.¹ Imaging modalities such as dermoscopy or RCM could help minimize the need for surgical biopsies. Dermoscopy, which has been proven to allow for more efficient and accurate diagnoses in dermatology, is reimbursed in Europe but not in the United States.^7-9 In 2016, CPT codes 96931 through 96936 were created for RCM and are covered by most insurances.¹⁰ Optical coherence tomography, another noninvasive imaging technology, currently is not reimbursed but did receive Category III codes (0470T-0471T), also known as a tracking codes, in 2017.⁴ Category III codes are used for emerging technologies that have future potential but do not have enough US-based evidence to support receiving Category I CPT codes. The use of Category III codes allows for data collection on emerging technologies and services, with the potential to convert the Category III codes to Category I codes once certain criteria are met.¹¹ 

Beginning in 2019, the standard biopsy codes 11100 and 11101 were replaced with 6 new codes to represent primary (11102, 11104, 11106) and add-on biopsies (11103, 11105, 11107) based on the sampling technique utilized and the thickness of the sample (Table 1). Previously, the biopsy codes did not reflect the complexity of the different biopsy techniques, whereas the new codes provide differentiation of the method of removal (ie, tangential, punch, incisional).^2,3 The base code is dependent on whichever biopsy performed has the highest complexity, with incisional biopsy--a partial excision--being considered the most complex.³ Punch biopsy is considered the next level of complexity, followed by tangential biopsy. Each of the 6 new biopsy codes also received a new wRVU, which determines reimbursement under Medicare and most other insurers when combined with direct peRVU and mRVU. Additional biopsies, reported using the add-on codes, are reimbursed at a lower level than the base codes because of removal of duplicate inputs for preservice and postservice care.³  

Telehealth Codes 

Telemedicine services offer another form of imaging that providers can use to communicate remotely with patients through a live interactive video stream (with audio), a store-and-forward system with photographs or videos shared asynchronously, or remote patient monitoring.¹² Although live video streaming uses a webcam, store-and-forward services involve sending photographs or videos electronically for later evaluation.^12,13 Remote patient monitoring allows the collection of health-related data and transmission to a physician without the need for an office visit.¹³ Most states require physicians to have a license in the state in which the patient is located at the time of the encounter. Given the difficulty of applying for licensure in multiple states, several states started creating their own special licenses to allow out-of-state providers to offer services through telemedicine.¹⁴ The Federation of State Medical Boards then created the Interstate Medical Licensure Compact (IMLC) for an expedited process to apply for medical licensure in other states. The IMLC was formed to increase access to health care in underserved or rural areas including but not limited to the use of telemedicine.¹⁵ To qualify for IMLC, a physician must have a medical license in a state registered with the IMLC (ie, state of principal license) and have at least one of the following in their state of principal license: primary residence, 25% of their medical practice, a current employer, or US federal income taxes filed.¹⁵ The remaining states that do not have a licensing process for telemedicine allow practice in contiguous states or may provide temporary licenses dependent on the situation.¹⁴ 

Since 2017, billing codes for telemedicine have been the same as those used for in-person evaluation and management services with modifiers -95 or GQ added to the end of the code. Modifier -95 has been used for real-time telemedicine services, while modifier GQ has been used for store-and-forward services.¹⁶ For example, the code 99201, which is used to bill for new patients at outpatient visits, would become 99201-95 if performed using a live audio and video feed or 99201-GQ if information was sent electronically for later analysis. To receive reimbursement from Medicare, modifier -95 requires real-time communication using both audio and video; however, modifier GQ is only reimbursable in federal telemedicine demonstration programs in Alaska or Hawaii.¹² Note that reimbursement is up to the discretion of private providers, and even Medicare reimbursement can vary from state to state. 

In 2019, new Healthcare Common Procedure Coding System telemedicine codes were introduced to include virtual check-ins (G2012) and evaluation of patient-transmitted images and videos (G2010). G2010 is the first store-and-forward code that has the potential to be reimbursed outside of Alaska or Hawaii.^3,12 G2012 allows providers to monitor the patients' well-being outside of the office setting, a cost-effective alternative if patients do not require a full visit. More detailed descriptions of the new codes can be found in Table 3.¹ 

Final Thoughts 

As insurance providers continue to better monitor health care costs, it is of utmost importance that physicians become more involved in accurately assessing their services and procedures, given that the changes in RVUs mirror the Centers for Medicare & Medicaid Services' utilization of the RUC's interpretation of our survey responses.¹ The current billing codes attempt to better represent the work involved for each service, one example being the modification to more specific biopsy codes in 2019.  

With the growth of technology, CPT and Healthcare Common Procedure Coding System codes also reflect a push toward more efficient health care delivery and broader coverage for provider services, as demonstrated by the introduction of new telemedicine codes as well as recent additions of noninvasive imaging codes. Although technology makes health care more cost-effective for patients, clinicians can still maintain their overall reimbursements by efficiently seeing an increasing number of patients; for example, a patient diagnosed noninvasively using RCM can then receive same-day care, which impacts patients' quality of life by minimizing travel time, number of office visits, and time taken off from work, while allowing providers to manage a higher patient volume more productively. The new CPT codes discussed here reflect the growth of medical technology potential, which increases our diagnostic capability, making it even more critical for physicians to engage with these developments. 
 

As the US population continues to grow and patients become more aware of their health needs, payers are beginning to recognize the benefits of more efficient and cost-effective health care. With the implementation of the new Medicare Physician Fee Schedule on January 1, 2019, some old billing codes were revalued while others were replaced entirely with new codes.¹ The restructuring of the standard biopsy codes now takes the complexity of different sampling techniques into consideration. Furthermore, Current Procedural Terminology (CPT) Category III tracking codes for some imaging devices (eg, optical coherence tomography) added in 2017 require more data before obtaining a Category I reimbursable code, while codes for other imaging devices such as reflectance confocal microscopy (RCM) remain relatively the same.^2-4 Notably, the majority of the new 2019 telemedicine codes are applicable to dermatology.^2,3 In this article, we discuss the new CPT codes for reporting diagnostic procedures, including biopsy, noninvasive imaging, and telemedicine services. We also provide a summary of the national average reimbursement rates for these procedures. 

Background on Reimbursement 

To better understand how reimbursement works, it is important to know that all billing codes are provided a relative value unit (RVU), a number representing the value of the work involved and cost of providing a service relative to other services.⁵ The total RVU consists of the work RVU (wRVU), practice expense RVU (peRVU), and malpractice expense RVU (mRVU). The wRVU represents the time, effort, and complexity involved in performing the service. The peRVU reflects the direct cost of supplies, personnel, and durable equipment involved in providing the service, excluding typical office overhead costs such as rent, utilities, and administrative staff. The mRVU is to cover the cost of malpractice insurance.⁵ The peRVU can be further specified as facility versus nonfacility services depending on where the service is performed.⁶ A facility peRVU is for services completed in a facility such as a hospital, outpatient hospital setting, or nursing home. The facility provides some of the involved supplies, personnel, and equipment for which they can recapture costs by separate reporting, resulting in a lower total RVU for the provider charges compared with nonfacility locations where the physician must provide these items.⁶ Many physicians may not be aware of how critical their role is in determining their own reimbursement rates by understanding RVUs and properly filling out Relative Value Scale Update Committee (RUC) surveys. If surveys sent to practitioners are accurately completed, RVUs have the potential to be fairly valued; however, if respondents are unaware of all of the components that are inherent to a procedure, they may end up minimizing the effort or time involved, which would skew the results and hurt those who perform the procedure. Rather than inputting appropriate preoperative and postoperative service times, many respondents often put 0s and 1s throughout the survey, which misrepresents the amount of time involved for a procedure. For example, inputting a preoperative time as 0 or 1 minute may severely underestimate the work involved for a procedure if the true preoperative time is 5 minutes. Such survey responses affect whether or not RVUs are valued appropriately. 

The billing codes and their RVUs as well as Medicare payment values in your area can be found on the Centers for Medicare & Medicaid Services website.^2,3 Table 1 provides a comparison of the old and new biopsy codes, and Table 2 shows the new RCM codes. 

Biopsy Codes 

Prior to 2019, biopsies were reimbursed using CPT code 11100 for the initial biopsy and 11101 for each additional biopsy.² Called up for refinement in the RUC process, initial data from the Physician Practice Expense Information Survey pointed to the likelihood of different sampling techniques having different amounts of work being supplied by different techniques.¹ Imaging modalities such as dermoscopy or RCM could help minimize the need for surgical biopsies. Dermoscopy, which has been proven to allow for more efficient and accurate diagnoses in dermatology, is reimbursed in Europe but not in the United States.^7-9 In 2016, CPT codes 96931 through 96936 were created for RCM and are covered by most insurances.¹⁰ Optical coherence tomography, another noninvasive imaging technology, currently is not reimbursed but did receive Category III codes (0470T-0471T), also known as a tracking codes, in 2017.⁴ Category III codes are used for emerging technologies that have future potential but do not have enough US-based evidence to support receiving Category I CPT codes. The use of Category III codes allows for data collection on emerging technologies and services, with the potential to convert the Category III codes to Category I codes once certain criteria are met.¹¹ 

Beginning in 2019, the standard biopsy codes 11100 and 11101 were replaced with 6 new codes to represent primary (11102, 11104, 11106) and add-on biopsies (11103, 11105, 11107) based on the sampling technique utilized and the thickness of the sample (Table 1). Previously, the biopsy codes did not reflect the complexity of the different biopsy techniques, whereas the new codes provide differentiation of the method of removal (ie, tangential, punch, incisional).^2,3 The base code is dependent on whichever biopsy performed has the highest complexity, with incisional biopsy--a partial excision--being considered the most complex.³ Punch biopsy is considered the next level of complexity, followed by tangential biopsy. Each of the 6 new biopsy codes also received a new wRVU, which determines reimbursement under Medicare and most other insurers when combined with direct peRVU and mRVU. Additional biopsies, reported using the add-on codes, are reimbursed at a lower level than the base codes because of removal of duplicate inputs for preservice and postservice care.³  

Telehealth Codes 

Telemedicine services offer another form of imaging that providers can use to communicate remotely with patients through a live interactive video stream (with audio), a store-and-forward system with photographs or videos shared asynchronously, or remote patient monitoring.¹² Although live video streaming uses a webcam, store-and-forward services involve sending photographs or videos electronically for later evaluation.^12,13 Remote patient monitoring allows the collection of health-related data and transmission to a physician without the need for an office visit.¹³ Most states require physicians to have a license in the state in which the patient is located at the time of the encounter. Given the difficulty of applying for licensure in multiple states, several states started creating their own special licenses to allow out-of-state providers to offer services through telemedicine.¹⁴ The Federation of State Medical Boards then created the Interstate Medical Licensure Compact (IMLC) for an expedited process to apply for medical licensure in other states. The IMLC was formed to increase access to health care in underserved or rural areas including but not limited to the use of telemedicine.¹⁵ To qualify for IMLC, a physician must have a medical license in a state registered with the IMLC (ie, state of principal license) and have at least one of the following in their state of principal license: primary residence, 25% of their medical practice, a current employer, or US federal income taxes filed.¹⁵ The remaining states that do not have a licensing process for telemedicine allow practice in contiguous states or may provide temporary licenses dependent on the situation.¹⁴ 

Since 2017, billing codes for telemedicine have been the same as those used for in-person evaluation and management services with modifiers -95 or GQ added to the end of the code. Modifier -95 has been used for real-time telemedicine services, while modifier GQ has been used for store-and-forward services.¹⁶ For example, the code 99201, which is used to bill for new patients at outpatient visits, would become 99201-95 if performed using a live audio and video feed or 99201-GQ if information was sent electronically for later analysis. To receive reimbursement from Medicare, modifier -95 requires real-time communication using both audio and video; however, modifier GQ is only reimbursable in federal telemedicine demonstration programs in Alaska or Hawaii.¹² Note that reimbursement is up to the discretion of private providers, and even Medicare reimbursement can vary from state to state. 

In 2019, new Healthcare Common Procedure Coding System telemedicine codes were introduced to include virtual check-ins (G2012) and evaluation of patient-transmitted images and videos (G2010). G2010 is the first store-and-forward code that has the potential to be reimbursed outside of Alaska or Hawaii.^3,12 G2012 allows providers to monitor the patients' well-being outside of the office setting, a cost-effective alternative if patients do not require a full visit. More detailed descriptions of the new codes can be found in Table 3.¹ 

Final Thoughts 

As insurance providers continue to better monitor health care costs, it is of utmost importance that physicians become more involved in accurately assessing their services and procedures, given that the changes in RVUs mirror the Centers for Medicare & Medicaid Services' utilization of the RUC's interpretation of our survey responses.¹ The current billing codes attempt to better represent the work involved for each service, one example being the modification to more specific biopsy codes in 2019.  

With the growth of technology, CPT and Healthcare Common Procedure Coding System codes also reflect a push toward more efficient health care delivery and broader coverage for provider services, as demonstrated by the introduction of new telemedicine codes as well as recent additions of noninvasive imaging codes. Although technology makes health care more cost-effective for patients, clinicians can still maintain their overall reimbursements by efficiently seeing an increasing number of patients; for example, a patient diagnosed noninvasively using RCM can then receive same-day care, which impacts patients' quality of life by minimizing travel time, number of office visits, and time taken off from work, while allowing providers to manage a higher patient volume more productively. The new CPT codes discussed here reflect the growth of medical technology potential, which increases our diagnostic capability, making it even more critical for physicians to engage with these developments. 
 

As technology continues to advance, so too does its accessibility to the general population. In 2013, 56% of Americans owned a smartphone versus 77% in 2017.¹With the increase in mobile applications (apps) available, it is no surprise that the market has extended into the medical field, with dermatology being no exception.² The majority of dermatology apps can be classified as teledermatology apps, followed by self-surveillance, disease guide, and reference apps. Additional types of dermatology apps include dermoscopy, conference, education, photograph storage and sharing, and journal apps, and others.² In this study, we examined Apple App Store rankings to determine the types of dermatology apps that are most popular among patients and physicians.

METHODS

A popular app rankings analyzer (App Annie) was used to search for dermatology apps along with their App Store rankings.³ Although iOS is not the most popular mobile device operating system, we chose to evaluate app rankings via the App Store because iPhones are the top-selling individual phones of any kind in the United States.⁴

We performed our analysis on a single day (July 14, 2018) given that app rankings can change daily. We incorporated the following keywords, which were commonly used in other dermatology app studies: dermatology, psoriasis, rosacea, acne, skin cancer, melanoma, eczema, and teledermatology. The category ranking was defined as the rank of a free or paid app in the App Store’s top charts for the selected country (United States), market (Apple), and device (iPhone) within their app category (Medical). Inclusion criteria required a ranking in the top 1500 Medical apps and being categorized in the App Store as a Medical app. Exclusion criteria included apps that focused on cosmetics, private practice, direct advertisements, photograph editing, or claims to cure skin disease, as well as non–English-language apps. The App Store descriptions were assessed to determine the type of each app (eg, teledermatology, disease guide) and target audience (patient, physician, or both).

Another search was performed using the same keywords but within the Health and Fitness category to capture potentially more highly ranked apps among patients. We also conducted separate searches within the Medical category using the keywords billing, coding, and ICD (International Classification of Diseases) to evaluate rankings for billing/coding apps, as well as EMR and electronic medical records for electronic medical record (EMR) apps.

RESULTS

The initial search yielded 851 results, which was narrowed down to 29 apps after applying the exclusion criteria. Of note, prior to application of the exclusion criteria, one dermatology app that was considered to be a direct advertisement app claiming to cure acne was ranked fourth of 1500 apps in the Medical category. However, the majority of the search results were excluded because they were not popular enough to be ranked among the top 1500 apps. There were more ranked dermatology apps in the Medical category targeting patients than physicians; 18 of 29 (62%) qualifying apps targeted patients and 11 (38%) targeted physicians (Tables 1 and 2). No apps targeted both groups. The most common type of ranked app targeting patients was self-surveillance (11/18), and the most common type targeting physicians was reference (8/11). The highest ranked app targeting patients was a teledermatology app with a ranking of 184, and the highest ranked app targeting physicians was educational, ranked 353. The least common type of ranked apps targeting patients were “other” (2/18 [11%]; 1 prescription and 1 UV monitor app) and conference (1/18 [6%]). The least common type of ranked apps targeting physicians were education (2/11 [18%]) and dermoscopy (1/11 [9%]).

Our search of the Health and Fitness category yielded 6 apps, all targeting patients; 3 (50%) were self-surveillance apps, and 3 (50%) were classified as other (2 UV monitors and a conferencing app for cancer emotional support)(Table 3).

Our search of the Medical category for billing/coding and EMR apps yielded 232 and 164 apps, respectively; of them, 49 (21%) and 54 (33%) apps were ranked. These apps did not overlap with the dermatology-related search criteria; thus, we were not able to ascertain how many of these apps were used specifically by health care providers in dermatology.

COMMENT

Patient Apps

The most common apps used by patients are fitness and nutrition tracker apps categorized as Health and Fitness^5,6; however, the majority of ranked dermatology apps are categorized as Medical per our findings. In a study of 557 dermatology patients, it was found that among the health-related apps they used, the most common apps after fitness/nutrition were references, followed by patient portals, self-surveillance, and emotional assistance apps.⁶ Our search was consistent with these findings, suggesting that the most desired dermatology apps by patients are those that allow them to be proactive with their health. It is no surprise that the top-ranked app targeting patients was a teledermatology app, followed by multiple self-surveillance apps. The highest ranked self-surveillance app in the Health and Fitness category focused on monitoring the effects of nutrition on symptoms of diseases including skin disorders, while the highest ranked (as well as the majority of) self-surveillance apps in the Medical category encompassed mole monitoring and cancer risk calculators.

Benefits of the ranked dermatology apps in the Medical and Health and Fitness categories targeting patients include more immediate access to health care and education. Despite this popularity among patients, Masud et al⁷ demonstrated that only 20.5% (9/44) of dermatology apps targeting patients may be reliable resources based on a rubric created by the investigators. Overall, there remains a research gap for a standardized scientific approach to evaluating app validity and reliability.

Teledermatology
Teledermatology apps are the most common dermatology apps,² allowing for remote evaluation of patients through either live consultations or transmittance of medical information for later review by board-certified physicians.⁸ Features common to many teledermatology apps include accessibility on Android (Google Inc) and iOS as well as a web version. Security and Health Insurance Portability and Accountability Act compliance is especially important and is enforced through user authentications, data encryption, and automatic logout features. Data is not stored locally and is secured on a private server with backup. Referring providers and consultants often can communicate within the app. Insurance providers also may cover teledermatology services, and if not, the out-of-pocket costs often are affordable.

The highest-ranked patient app (ranked 184 in the Medical category) was a teledermatology app that did not meet the American Telemedicine Association standards for teledermatology apps.⁹ The popularity of this app among patients may have been attributable to multiple ease-of-use and turnaround time features. The user interface was simplistic, and the design was appealing to the eye. The entry field options were minimal to avoid confusion. The turnaround time to receive a diagnosis depended on 1 of 3 options, including a more rapid response for an increased cost. Ease of use was the highlight of this app at the cost of accuracy, as the limited amount of information that users were required to provide physicians compromised diagnostic accuracy in this app.

For comparison, we chose a nonranked (and thus less frequently used) teledermatology app that had previously undergone scientific evaluation using 13 evaluation criteria specific to teledermatology.¹⁰ The app also met the American Telemedicine Association standard for teledermatology apps.⁹ The app was originally a broader telemedicine app but featured a section specific to teledermatology. The user interface was simple but professional, almost resembling an EMR. The input fields included a comprehensive history that permitted a better evaluation of a lesion but might be tedious for users. This app boasted professionalism and accuracy, but from a user standpoint, it may have been too time-consuming.

Striking a balance between ensuring proper care versus appealing to patients is a difficult but important task. Based on this study, it appears that popular patient apps may in fact have less scientific rationale and therefore potentially less accuracy.

Self-surveillance
Although self-surveillance apps did not account for the highest-ranked app, they were the most frequently ranked app type in our study. Most of the ranked self-surveillance apps in the Medical category were for monitoring lesions over time to assess for changes. These apps help users take photographs that are well organized in a single, easy-to-find location. Some apps were risk calculators that assessed the risk for malignancies using a questionnaire. The majority of these self-surveillance apps were specific to skin cancer detection. Of note, one of the ranked self-surveillance apps assessed drug effectiveness by monitoring clinical appearance and symptoms. The lowest ranked self-surveillance app in the top 1500 ranked Medical apps in our search monitored cancer symptoms not specific to dermatology. Although this app had a low ranking (1380/1500), it received a high number of reviews and was well rated at 4.8 out of 5 stars; therefore, it seemed more helpful than the other higher-ranked apps targeting patients, which had higher rankings but minimal to no reviews or ratings. A comparison of the ease-of-use features of all the ranked patient-targeted self-surveillance apps in the Medical category is provided in Table 4.

Physician Apps

After examining the results of apps targeting physicians, we realized that the data may be accurate but may not be as representative of all currently practicing dermatology providers. Given the increased usage of apps among younger age groups,¹¹ our data may be skewed toward medical students and residents, supported by the fact that the top-ranked physician app in our study was an education app and the majority were reference apps. Future studies are needed to reexamine app ranking as this age group transitions from entry-level health care providers in the next 5 to 10 years. These findings also suggest less frequent app use among more veteran health care providers within our specific search parameters. Therefore, we decided to do subsequent searches for available billing/coding and EMR apps, which were many, but as mentioned above, none were specific to dermatology.

General Dermatology References
Most of the dermatology reference apps were formatted as e-books; however, other apps such as the Amazon Kindle app (categorized under Books) providing access to multiple e-books within one app were not included. Some apps included study aid features (eg, flash cards, quizzes), and topics spanned both dermatology and dermatopathology. Apps provide a unique way for on-the-go studying for dermatologists in training, and if the usage continues to grow, there may be a need for increased formal integration in dermatology education in the future.

Journals
Journal apps were not among those listed in the top-ranked apps we evaluated, which we suspect may be because journals were categorized differently from one journal to the next; for example, the Journal of the American Academy of Dermatology was ranked 1168 in the Magazines and Newspapers category. On the other hand, Dermatology World was ranked 1363 in the Reference category. An article’s citation affects the publishing journal’s impact factor, which is one of the most important variables in measuring a journal’s influence. In the future, there may be other variables that could aid in understanding journal impact as it relates to the journal’s accessibility.

Limitations

Our study did not look at Android apps. The top chart apps in the Android and Apple App Stores use undisclosed algorithms likely involving different characteristics such as number of downloads, frequency of updates, number of reviews, ratings, and more. Thus, the rankings across these different markets would not be comparable. Although our choice of keywords stemmed from the majority of prior studies looking at dermatology apps, our search was limited due to the use of these specific keywords. To avoid skewing data by cross-comparison of noncomparable categories, we could not compare apps in the Medical category versus those in other categories.

CONCLUSION

There seems to be a disconnect between the apps that are popular among patients and the scientific validity of the apps. As app usage increases among dermatology providers, whose demographic is shifting younger and younger, apps may become more incorporated in our education, and as such, it will become more critical to develop formal scientific standards. Given these future trends, we may need to increase our current literature and understanding of apps in dermatology with regard to their impact on both patients and health care providers.

As technology continues to advance, so too does its accessibility to the general population. In 2013, 56% of Americans owned a smartphone versus 77% in 2017.¹With the increase in mobile applications (apps) available, it is no surprise that the market has extended into the medical field, with dermatology being no exception.² The majority of dermatology apps can be classified as teledermatology apps, followed by self-surveillance, disease guide, and reference apps. Additional types of dermatology apps include dermoscopy, conference, education, photograph storage and sharing, and journal apps, and others.² In this study, we examined Apple App Store rankings to determine the types of dermatology apps that are most popular among patients and physicians.

METHODS

A popular app rankings analyzer (App Annie) was used to search for dermatology apps along with their App Store rankings.³ Although iOS is not the most popular mobile device operating system, we chose to evaluate app rankings via the App Store because iPhones are the top-selling individual phones of any kind in the United States.⁴

We performed our analysis on a single day (July 14, 2018) given that app rankings can change daily. We incorporated the following keywords, which were commonly used in other dermatology app studies: dermatology, psoriasis, rosacea, acne, skin cancer, melanoma, eczema, and teledermatology. The category ranking was defined as the rank of a free or paid app in the App Store’s top charts for the selected country (United States), market (Apple), and device (iPhone) within their app category (Medical). Inclusion criteria required a ranking in the top 1500 Medical apps and being categorized in the App Store as a Medical app. Exclusion criteria included apps that focused on cosmetics, private practice, direct advertisements, photograph editing, or claims to cure skin disease, as well as non–English-language apps. The App Store descriptions were assessed to determine the type of each app (eg, teledermatology, disease guide) and target audience (patient, physician, or both).

Another search was performed using the same keywords but within the Health and Fitness category to capture potentially more highly ranked apps among patients. We also conducted separate searches within the Medical category using the keywords billing, coding, and ICD (International Classification of Diseases) to evaluate rankings for billing/coding apps, as well as EMR and electronic medical records for electronic medical record (EMR) apps.

RESULTS

The initial search yielded 851 results, which was narrowed down to 29 apps after applying the exclusion criteria. Of note, prior to application of the exclusion criteria, one dermatology app that was considered to be a direct advertisement app claiming to cure acne was ranked fourth of 1500 apps in the Medical category. However, the majority of the search results were excluded because they were not popular enough to be ranked among the top 1500 apps. There were more ranked dermatology apps in the Medical category targeting patients than physicians; 18 of 29 (62%) qualifying apps targeted patients and 11 (38%) targeted physicians (Tables 1 and 2). No apps targeted both groups. The most common type of ranked app targeting patients was self-surveillance (11/18), and the most common type targeting physicians was reference (8/11). The highest ranked app targeting patients was a teledermatology app with a ranking of 184, and the highest ranked app targeting physicians was educational, ranked 353. The least common type of ranked apps targeting patients were “other” (2/18 [11%]; 1 prescription and 1 UV monitor app) and conference (1/18 [6%]). The least common type of ranked apps targeting physicians were education (2/11 [18%]) and dermoscopy (1/11 [9%]).

Our search of the Health and Fitness category yielded 6 apps, all targeting patients; 3 (50%) were self-surveillance apps, and 3 (50%) were classified as other (2 UV monitors and a conferencing app for cancer emotional support)(Table 3).

Our search of the Medical category for billing/coding and EMR apps yielded 232 and 164 apps, respectively; of them, 49 (21%) and 54 (33%) apps were ranked. These apps did not overlap with the dermatology-related search criteria; thus, we were not able to ascertain how many of these apps were used specifically by health care providers in dermatology.

COMMENT

Patient Apps

The most common apps used by patients are fitness and nutrition tracker apps categorized as Health and Fitness^5,6; however, the majority of ranked dermatology apps are categorized as Medical per our findings. In a study of 557 dermatology patients, it was found that among the health-related apps they used, the most common apps after fitness/nutrition were references, followed by patient portals, self-surveillance, and emotional assistance apps.⁶ Our search was consistent with these findings, suggesting that the most desired dermatology apps by patients are those that allow them to be proactive with their health. It is no surprise that the top-ranked app targeting patients was a teledermatology app, followed by multiple self-surveillance apps. The highest ranked self-surveillance app in the Health and Fitness category focused on monitoring the effects of nutrition on symptoms of diseases including skin disorders, while the highest ranked (as well as the majority of) self-surveillance apps in the Medical category encompassed mole monitoring and cancer risk calculators.

Benefits of the ranked dermatology apps in the Medical and Health and Fitness categories targeting patients include more immediate access to health care and education. Despite this popularity among patients, Masud et al⁷ demonstrated that only 20.5% (9/44) of dermatology apps targeting patients may be reliable resources based on a rubric created by the investigators. Overall, there remains a research gap for a standardized scientific approach to evaluating app validity and reliability.

Teledermatology
Teledermatology apps are the most common dermatology apps,² allowing for remote evaluation of patients through either live consultations or transmittance of medical information for later review by board-certified physicians.⁸ Features common to many teledermatology apps include accessibility on Android (Google Inc) and iOS as well as a web version. Security and Health Insurance Portability and Accountability Act compliance is especially important and is enforced through user authentications, data encryption, and automatic logout features. Data is not stored locally and is secured on a private server with backup. Referring providers and consultants often can communicate within the app. Insurance providers also may cover teledermatology services, and if not, the out-of-pocket costs often are affordable.

The highest-ranked patient app (ranked 184 in the Medical category) was a teledermatology app that did not meet the American Telemedicine Association standards for teledermatology apps.⁹ The popularity of this app among patients may have been attributable to multiple ease-of-use and turnaround time features. The user interface was simplistic, and the design was appealing to the eye. The entry field options were minimal to avoid confusion. The turnaround time to receive a diagnosis depended on 1 of 3 options, including a more rapid response for an increased cost. Ease of use was the highlight of this app at the cost of accuracy, as the limited amount of information that users were required to provide physicians compromised diagnostic accuracy in this app.

For comparison, we chose a nonranked (and thus less frequently used) teledermatology app that had previously undergone scientific evaluation using 13 evaluation criteria specific to teledermatology.¹⁰ The app also met the American Telemedicine Association standard for teledermatology apps.⁹ The app was originally a broader telemedicine app but featured a section specific to teledermatology. The user interface was simple but professional, almost resembling an EMR. The input fields included a comprehensive history that permitted a better evaluation of a lesion but might be tedious for users. This app boasted professionalism and accuracy, but from a user standpoint, it may have been too time-consuming.

Striking a balance between ensuring proper care versus appealing to patients is a difficult but important task. Based on this study, it appears that popular patient apps may in fact have less scientific rationale and therefore potentially less accuracy.

Self-surveillance
Although self-surveillance apps did not account for the highest-ranked app, they were the most frequently ranked app type in our study. Most of the ranked self-surveillance apps in the Medical category were for monitoring lesions over time to assess for changes. These apps help users take photographs that are well organized in a single, easy-to-find location. Some apps were risk calculators that assessed the risk for malignancies using a questionnaire. The majority of these self-surveillance apps were specific to skin cancer detection. Of note, one of the ranked self-surveillance apps assessed drug effectiveness by monitoring clinical appearance and symptoms. The lowest ranked self-surveillance app in the top 1500 ranked Medical apps in our search monitored cancer symptoms not specific to dermatology. Although this app had a low ranking (1380/1500), it received a high number of reviews and was well rated at 4.8 out of 5 stars; therefore, it seemed more helpful than the other higher-ranked apps targeting patients, which had higher rankings but minimal to no reviews or ratings. A comparison of the ease-of-use features of all the ranked patient-targeted self-surveillance apps in the Medical category is provided in Table 4.

Physician Apps

After examining the results of apps targeting physicians, we realized that the data may be accurate but may not be as representative of all currently practicing dermatology providers. Given the increased usage of apps among younger age groups,¹¹ our data may be skewed toward medical students and residents, supported by the fact that the top-ranked physician app in our study was an education app and the majority were reference apps. Future studies are needed to reexamine app ranking as this age group transitions from entry-level health care providers in the next 5 to 10 years. These findings also suggest less frequent app use among more veteran health care providers within our specific search parameters. Therefore, we decided to do subsequent searches for available billing/coding and EMR apps, which were many, but as mentioned above, none were specific to dermatology.

General Dermatology References
Most of the dermatology reference apps were formatted as e-books; however, other apps such as the Amazon Kindle app (categorized under Books) providing access to multiple e-books within one app were not included. Some apps included study aid features (eg, flash cards, quizzes), and topics spanned both dermatology and dermatopathology. Apps provide a unique way for on-the-go studying for dermatologists in training, and if the usage continues to grow, there may be a need for increased formal integration in dermatology education in the future.

Journals
Journal apps were not among those listed in the top-ranked apps we evaluated, which we suspect may be because journals were categorized differently from one journal to the next; for example, the Journal of the American Academy of Dermatology was ranked 1168 in the Magazines and Newspapers category. On the other hand, Dermatology World was ranked 1363 in the Reference category. An article’s citation affects the publishing journal’s impact factor, which is one of the most important variables in measuring a journal’s influence. In the future, there may be other variables that could aid in understanding journal impact as it relates to the journal’s accessibility.

Limitations

Our study did not look at Android apps. The top chart apps in the Android and Apple App Stores use undisclosed algorithms likely involving different characteristics such as number of downloads, frequency of updates, number of reviews, ratings, and more. Thus, the rankings across these different markets would not be comparable. Although our choice of keywords stemmed from the majority of prior studies looking at dermatology apps, our search was limited due to the use of these specific keywords. To avoid skewing data by cross-comparison of noncomparable categories, we could not compare apps in the Medical category versus those in other categories.

CONCLUSION

There seems to be a disconnect between the apps that are popular among patients and the scientific validity of the apps. As app usage increases among dermatology providers, whose demographic is shifting younger and younger, apps may become more incorporated in our education, and as such, it will become more critical to develop formal scientific standards. Given these future trends, we may need to increase our current literature and understanding of apps in dermatology with regard to their impact on both patients and health care providers.

As technology continues to advance, so too does its accessibility to the general population. In 2013, 56% of Americans owned a smartphone versus 77% in 2017.¹With the increase in mobile applications (apps) available, it is no surprise that the market has extended into the medical field, with dermatology being no exception.² The majority of dermatology apps can be classified as teledermatology apps, followed by self-surveillance, disease guide, and reference apps. Additional types of dermatology apps include dermoscopy, conference, education, photograph storage and sharing, and journal apps, and others.² In this study, we examined Apple App Store rankings to determine the types of dermatology apps that are most popular among patients and physicians.

METHODS

A popular app rankings analyzer (App Annie) was used to search for dermatology apps along with their App Store rankings.³ Although iOS is not the most popular mobile device operating system, we chose to evaluate app rankings via the App Store because iPhones are the top-selling individual phones of any kind in the United States.⁴

We performed our analysis on a single day (July 14, 2018) given that app rankings can change daily. We incorporated the following keywords, which were commonly used in other dermatology app studies: dermatology, psoriasis, rosacea, acne, skin cancer, melanoma, eczema, and teledermatology. The category ranking was defined as the rank of a free or paid app in the App Store’s top charts for the selected country (United States), market (Apple), and device (iPhone) within their app category (Medical). Inclusion criteria required a ranking in the top 1500 Medical apps and being categorized in the App Store as a Medical app. Exclusion criteria included apps that focused on cosmetics, private practice, direct advertisements, photograph editing, or claims to cure skin disease, as well as non–English-language apps. The App Store descriptions were assessed to determine the type of each app (eg, teledermatology, disease guide) and target audience (patient, physician, or both).

Another search was performed using the same keywords but within the Health and Fitness category to capture potentially more highly ranked apps among patients. We also conducted separate searches within the Medical category using the keywords billing, coding, and ICD (International Classification of Diseases) to evaluate rankings for billing/coding apps, as well as EMR and electronic medical records for electronic medical record (EMR) apps.

RESULTS

The initial search yielded 851 results, which was narrowed down to 29 apps after applying the exclusion criteria. Of note, prior to application of the exclusion criteria, one dermatology app that was considered to be a direct advertisement app claiming to cure acne was ranked fourth of 1500 apps in the Medical category. However, the majority of the search results were excluded because they were not popular enough to be ranked among the top 1500 apps. There were more ranked dermatology apps in the Medical category targeting patients than physicians; 18 of 29 (62%) qualifying apps targeted patients and 11 (38%) targeted physicians (Tables 1 and 2). No apps targeted both groups. The most common type of ranked app targeting patients was self-surveillance (11/18), and the most common type targeting physicians was reference (8/11). The highest ranked app targeting patients was a teledermatology app with a ranking of 184, and the highest ranked app targeting physicians was educational, ranked 353. The least common type of ranked apps targeting patients were “other” (2/18 [11%]; 1 prescription and 1 UV monitor app) and conference (1/18 [6%]). The least common type of ranked apps targeting physicians were education (2/11 [18%]) and dermoscopy (1/11 [9%]).

Our search of the Health and Fitness category yielded 6 apps, all targeting patients; 3 (50%) were self-surveillance apps, and 3 (50%) were classified as other (2 UV monitors and a conferencing app for cancer emotional support)(Table 3).

Our search of the Medical category for billing/coding and EMR apps yielded 232 and 164 apps, respectively; of them, 49 (21%) and 54 (33%) apps were ranked. These apps did not overlap with the dermatology-related search criteria; thus, we were not able to ascertain how many of these apps were used specifically by health care providers in dermatology.

COMMENT

Patient Apps

The most common apps used by patients are fitness and nutrition tracker apps categorized as Health and Fitness^5,6; however, the majority of ranked dermatology apps are categorized as Medical per our findings. In a study of 557 dermatology patients, it was found that among the health-related apps they used, the most common apps after fitness/nutrition were references, followed by patient portals, self-surveillance, and emotional assistance apps.⁶ Our search was consistent with these findings, suggesting that the most desired dermatology apps by patients are those that allow them to be proactive with their health. It is no surprise that the top-ranked app targeting patients was a teledermatology app, followed by multiple self-surveillance apps. The highest ranked self-surveillance app in the Health and Fitness category focused on monitoring the effects of nutrition on symptoms of diseases including skin disorders, while the highest ranked (as well as the majority of) self-surveillance apps in the Medical category encompassed mole monitoring and cancer risk calculators.

Benefits of the ranked dermatology apps in the Medical and Health and Fitness categories targeting patients include more immediate access to health care and education. Despite this popularity among patients, Masud et al⁷ demonstrated that only 20.5% (9/44) of dermatology apps targeting patients may be reliable resources based on a rubric created by the investigators. Overall, there remains a research gap for a standardized scientific approach to evaluating app validity and reliability.

Teledermatology
Teledermatology apps are the most common dermatology apps,² allowing for remote evaluation of patients through either live consultations or transmittance of medical information for later review by board-certified physicians.⁸ Features common to many teledermatology apps include accessibility on Android (Google Inc) and iOS as well as a web version. Security and Health Insurance Portability and Accountability Act compliance is especially important and is enforced through user authentications, data encryption, and automatic logout features. Data is not stored locally and is secured on a private server with backup. Referring providers and consultants often can communicate within the app. Insurance providers also may cover teledermatology services, and if not, the out-of-pocket costs often are affordable.

The highest-ranked patient app (ranked 184 in the Medical category) was a teledermatology app that did not meet the American Telemedicine Association standards for teledermatology apps.⁹ The popularity of this app among patients may have been attributable to multiple ease-of-use and turnaround time features. The user interface was simplistic, and the design was appealing to the eye. The entry field options were minimal to avoid confusion. The turnaround time to receive a diagnosis depended on 1 of 3 options, including a more rapid response for an increased cost. Ease of use was the highlight of this app at the cost of accuracy, as the limited amount of information that users were required to provide physicians compromised diagnostic accuracy in this app.

For comparison, we chose a nonranked (and thus less frequently used) teledermatology app that had previously undergone scientific evaluation using 13 evaluation criteria specific to teledermatology.¹⁰ The app also met the American Telemedicine Association standard for teledermatology apps.⁹ The app was originally a broader telemedicine app but featured a section specific to teledermatology. The user interface was simple but professional, almost resembling an EMR. The input fields included a comprehensive history that permitted a better evaluation of a lesion but might be tedious for users. This app boasted professionalism and accuracy, but from a user standpoint, it may have been too time-consuming.

Striking a balance between ensuring proper care versus appealing to patients is a difficult but important task. Based on this study, it appears that popular patient apps may in fact have less scientific rationale and therefore potentially less accuracy.

Self-surveillance
Although self-surveillance apps did not account for the highest-ranked app, they were the most frequently ranked app type in our study. Most of the ranked self-surveillance apps in the Medical category were for monitoring lesions over time to assess for changes. These apps help users take photographs that are well organized in a single, easy-to-find location. Some apps were risk calculators that assessed the risk for malignancies using a questionnaire. The majority of these self-surveillance apps were specific to skin cancer detection. Of note, one of the ranked self-surveillance apps assessed drug effectiveness by monitoring clinical appearance and symptoms. The lowest ranked self-surveillance app in the top 1500 ranked Medical apps in our search monitored cancer symptoms not specific to dermatology. Although this app had a low ranking (1380/1500), it received a high number of reviews and was well rated at 4.8 out of 5 stars; therefore, it seemed more helpful than the other higher-ranked apps targeting patients, which had higher rankings but minimal to no reviews or ratings. A comparison of the ease-of-use features of all the ranked patient-targeted self-surveillance apps in the Medical category is provided in Table 4.

Physician Apps

After examining the results of apps targeting physicians, we realized that the data may be accurate but may not be as representative of all currently practicing dermatology providers. Given the increased usage of apps among younger age groups,¹¹ our data may be skewed toward medical students and residents, supported by the fact that the top-ranked physician app in our study was an education app and the majority were reference apps. Future studies are needed to reexamine app ranking as this age group transitions from entry-level health care providers in the next 5 to 10 years. These findings also suggest less frequent app use among more veteran health care providers within our specific search parameters. Therefore, we decided to do subsequent searches for available billing/coding and EMR apps, which were many, but as mentioned above, none were specific to dermatology.

General Dermatology References
Most of the dermatology reference apps were formatted as e-books; however, other apps such as the Amazon Kindle app (categorized under Books) providing access to multiple e-books within one app were not included. Some apps included study aid features (eg, flash cards, quizzes), and topics spanned both dermatology and dermatopathology. Apps provide a unique way for on-the-go studying for dermatologists in training, and if the usage continues to grow, there may be a need for increased formal integration in dermatology education in the future.

Journals
Journal apps were not among those listed in the top-ranked apps we evaluated, which we suspect may be because journals were categorized differently from one journal to the next; for example, the Journal of the American Academy of Dermatology was ranked 1168 in the Magazines and Newspapers category. On the other hand, Dermatology World was ranked 1363 in the Reference category. An article’s citation affects the publishing journal’s impact factor, which is one of the most important variables in measuring a journal’s influence. In the future, there may be other variables that could aid in understanding journal impact as it relates to the journal’s accessibility.

Limitations

Our study did not look at Android apps. The top chart apps in the Android and Apple App Stores use undisclosed algorithms likely involving different characteristics such as number of downloads, frequency of updates, number of reviews, ratings, and more. Thus, the rankings across these different markets would not be comparable. Although our choice of keywords stemmed from the majority of prior studies looking at dermatology apps, our search was limited due to the use of these specific keywords. To avoid skewing data by cross-comparison of noncomparable categories, we could not compare apps in the Medical category versus those in other categories.

CONCLUSION

There seems to be a disconnect between the apps that are popular among patients and the scientific validity of the apps. As app usage increases among dermatology providers, whose demographic is shifting younger and younger, apps may become more incorporated in our education, and as such, it will become more critical to develop formal scientific standards. Given these future trends, we may need to increase our current literature and understanding of apps in dermatology with regard to their impact on both patients and health care providers.

Dermoscopic examination has been proven to increase diagnostic accuracy and decrease unnecessary biopsies of both melanoma and nonmelanoma skin cancers.^1,2 Digital dermoscopy refers to acquiring and storing digital dermoscopic photographs via digital camera, smart image capture devices such as smartphones and tablets, or any other devices used for image acquisition. The stored images may then be used in a variety of ways, including sequential digital monitoring, teledermoscopy, and machine learning.

Sequential Digital Monitoring

Sequential digital dermoscopy imaging (SDDI) is the capture and storage of dermoscopic images of suspicious lesions that are then monitored over time for changes. Studies have shown that SDDI allows for early detection of melanomas and leads to a decrease in the number of unnecessary excisions.^3,4 A meta-analysis of SDDI found that the chance of detecting melanoma increased with the length of monitoring, which suggests that continued follow-up, especially in high-risk groups, is crucial.⁴

Teledermoscopy

Teledermatology (telederm) is on the rise in the United States, with the number of programs and consultations increasing yearly. One study showed a 48% increase in telederm programs in the last 5 years.⁵ Studies have shown the addition of digital dermoscopic images improved the diagnostic accuracy in telederm skin cancer screenings versus clinical images alone.^6,7

Telederm currently is practiced in 2 main models: live-interactive video consultation and storage of images for future consultation (store and forward). Medicare currently only reimburses live-interactive telederm for patients in nonmetropolitan areas and store-and-forward telederm pilot programs in Alaska and Hawaii; however, Medicaid does reimburse for store and forward in a handful of states.8 Similar to dermatoscope use during clinical examination, there currently is no additional reimbursement for teledermoscopy. Of note, a willingness-to-pay survey of 214 students from a southwestern university health center showed that participants were willing to pay an average (SD) of $55.27 ($39.11) out of pocket for a teledermoscopy/telederm evaluation, citing factors such as convenience.⁹

Direct-to-consumer telederm offers a new way for patients to receive care.¹⁰ Some dermatoscopes (eg, DermLite HÜD [3Gen], Molescope/Molescope II [Metaoptima Technology Inc]) currently are marketed directly to consumers along with telederm services to facilitate direct-to-patient teledermoscopy.^11,12

Machine Learning

Big data and machine learning has been hailed as the future of medicine and dermatology alike.¹³ Machine learning is a type of artificial intelligence that uses computational algorithms (eg, neural networks) that allow computer programs to automatically improve their accuracy (learn) by analyzing large data sets. In dermatology, machine learning has been most notably used to train computers to identify images of skin cancer by way of large image databases.^14-17 One algorithm, a convolutional neural network (CNN), made headlines in 2017 when it was able to identify dermoscopic and clinical images of skin cancer with comparable accuracy to a group of 21 dermatologists.¹⁴ In 2018, the International Skin Imaging Collaboration (ISIC) published results of a study of the diagnostic accuracy of 25 computer algorithms compared to 8 dermatologists using a set of 100 dermoscopic images of melanoma and benign nevi.¹⁵ Using the average sensitivity of the dermatologists (82%), the top fusion algorithm in the study had a sensitivity of 76% versus 59% for the dermatologists (P=.02). These results compared the mean sensitivity of the dermatologists, as some individual dermatologists outperformed the algorithm.¹⁵ More recently, another CNN was compared to 58 international dermatologists in the classification of a set of 100 dermoscopic images (20 melanoma and 80 melanocytic nevi).¹⁶ Using the mean sensitivity of the dermatologists (86.6%), the CNN had a specificity of 92.5% versus 71.3% for dermatologists (P<.01). In the second part of the study, the dermatologists were given some clinical information and close-up photographs of the lesions, which improved their average (SD) sensitivity and specificity to 88.9% (9.6%)(P=.19) and 75.7% (11.7%)(P<.05), respectively. When compared to the CNN at this higher sensitivity, the CNN still had a higher specificity than the dermatologists (82.5% vs 75.7% [P<.01]).¹⁶ However, in real-life clinical practice dermatologists perform better, not only because they can collect more in-person clinical information but also because humans gather more information during live examination than when they are interpreting close-up clinical and/or dermoscopic images. In a sense, we currently are limited to comparing data that is incommensurable.

Machine learning studies have other notable limitations, such as data sets that do not contain a full spectrum of skin lesions or less common lesions (eg, pigmented seborrheic keratoses, amelanotic melanomas) and variation in image databases used.^15,16 For machine algorithms to improve, they require access to high-quality and ideally standardized digital dermoscopic image databases. The ISIC and other organizations currently have databases specifically for this purpose, but more images are needed.¹⁸ As additional practitioners incorporate digital dermoscopy in their clinical practice, the potential for larger databases and more accurate algorithms becomes a possibility. 

Image Acquisition

Many devices are available for digital dermoscopic image acquisition, including dermatoscopes that attach to smartphones and/or digital cameras and all-in-one systems (eTable). The exact system employed will depend on the practitioner's requirements for price, portability, speed, image quality, and software. Digital single-lens reflex (DSLR) cameras boast the highest image quality, while video dermoscopy traditionally yields stored images with poor resolution.¹⁹ Macroscopic images obtained by other imaging devices, including spectral imaging devices and reflectance confocal microscopy, usually are yielded via video dermoscopy or a video camera to capture images; thus, stored images generally are not as high quality. 

Smartphones are increasingly used for clinical imaging in dermatology.²⁰ Although DSLR cameras still take the highest-quality images, current smartphone image quality is comparable to digital cameras.^21,22 Computational photography uses computer processing power to enhance image quality and may bring smartphone image quality closer to DSLR cameras.^22,23 Smartphones with newer dual-lens cameras have been reported to further improve image quality.²¹ Current smartphones have the option of enabling high-dynamic-range imaging, which combines multiple images taken with different exposures to create a single image with improved dynamic range of luminosity. It has been reported that high-dynamic-range imaging may even enhance dermoscopic features of more challenging hypopigmented skin cancers.²⁴

Standardizing Imaging

There has been a concerted effort to standardize digital dermatologic image acquisition.^25,26 Standardization promises to facilitate data analysis, improve collaboration, protect patient privacy, and improve patient care.^13,26,27 At the forefront of image standardization is the ISIC organization, which recently published its Delphi consensus guidelines on standards for lesion imaging, including dermoscopy.²⁶

The true holy grail of image standardization is the Digital Imaging and Communications in Medicine (DICOM) standard.^26-28 The DICOM is a comprehensive imaging standard for storage, annotation, transfer, and display of images, and it is most notable for its use in radiology. The DICOM also could be applied to new imaging modalities in dermatology (eg, optical coherence tomography, reflectance confocal microscopy). Past efforts to develop a DICOM standard for dermatology were undertaken by a working group that has since disbanded.²⁷ Work by the ISIC and many others will hopefully lead to adoption of the DICOM standard by dermatology at some point in the future. 

Protected Health Information

The Health Insurance Portability and Accountability Act (HIPAA) requires protected health information (PHI) to be stored in a secure manner with limited access that sufficiently protects identifiable patient information. Although dermoscopic images generally are deidentified, they often are stored alongside clinical photographs and data that contains PHI in clinical practice.

Image storage can take 2 forms: (1) physical local storage on internal and external hard drives or (2) remote storage (eg, cloud-based storage). Encryption is essential regardless of the method of storage. It is required by law that loss of nonencrypted PHI be reported to all potentially affected patients, the US Department of Health & Human Services, and local/state media depending on the number of patients affected. Loss of PHI can result in fines of up to $1.5 million.²⁹ On the contrary, loss of properly encrypted data would not be required to be reported.³⁰

As smart image acquisition devices begin to dominate the clinical setting, practitioners need to be vigilant in securing patient PHI. There are multiple applications (apps) that allow for secure encrypted digital dermoscopic image acquisition and storage on smartphones. Additionally, it is important to secure smartphones with complex passcodes (eg, a mix of special characters, numbers, uppercase and lowercase letters). Most dermatoscope manufacturers have apps for image acquisition and storage that can be tied into other platforms or storage systems (eg, DermLite app [3Gen], Handyscope [FotoFinder Systems GmbH], VEOS app [Canfield Scientific, Inc]).²⁸ Other options include syncing images with current electronic medical record technologies, transferring photographs to HIPAA-compliant cloud storage, or transferring photographs to an encrypted computer and/or external hard drive. Some tips for securing data based on HIPAA and other guidelines are listed in the Table.^30,31

Conclusion

The expansion of teledermoscopy alongside direct-to-patient services may create additional incentives for clinicians to incorporate digital dermoscopy into their practice. As more practitioners adopt digital dermoscopy, machine learning driven by technological advancements and larger image data sets could influence the future practice of dermatology. With the rise in digital dermoscopy by way of smartphones, additional steps must be taken to ensure patients' PHI is safeguarded. Digital dermoscopy is a dynamic field that will likely see continued growth in the coming years.

Dermoscopic examination has been proven to increase diagnostic accuracy and decrease unnecessary biopsies of both melanoma and nonmelanoma skin cancers.^1,2 Digital dermoscopy refers to acquiring and storing digital dermoscopic photographs via digital camera, smart image capture devices such as smartphones and tablets, or any other devices used for image acquisition. The stored images may then be used in a variety of ways, including sequential digital monitoring, teledermoscopy, and machine learning.

Sequential Digital Monitoring

Sequential digital dermoscopy imaging (SDDI) is the capture and storage of dermoscopic images of suspicious lesions that are then monitored over time for changes. Studies have shown that SDDI allows for early detection of melanomas and leads to a decrease in the number of unnecessary excisions.^3,4 A meta-analysis of SDDI found that the chance of detecting melanoma increased with the length of monitoring, which suggests that continued follow-up, especially in high-risk groups, is crucial.⁴

Teledermoscopy

Teledermatology (telederm) is on the rise in the United States, with the number of programs and consultations increasing yearly. One study showed a 48% increase in telederm programs in the last 5 years.⁵ Studies have shown the addition of digital dermoscopic images improved the diagnostic accuracy in telederm skin cancer screenings versus clinical images alone.^6,7

Telederm currently is practiced in 2 main models: live-interactive video consultation and storage of images for future consultation (store and forward). Medicare currently only reimburses live-interactive telederm for patients in nonmetropolitan areas and store-and-forward telederm pilot programs in Alaska and Hawaii; however, Medicaid does reimburse for store and forward in a handful of states.8 Similar to dermatoscope use during clinical examination, there currently is no additional reimbursement for teledermoscopy. Of note, a willingness-to-pay survey of 214 students from a southwestern university health center showed that participants were willing to pay an average (SD) of $55.27 ($39.11) out of pocket for a teledermoscopy/telederm evaluation, citing factors such as convenience.⁹

Direct-to-consumer telederm offers a new way for patients to receive care.¹⁰ Some dermatoscopes (eg, DermLite HÜD [3Gen], Molescope/Molescope II [Metaoptima Technology Inc]) currently are marketed directly to consumers along with telederm services to facilitate direct-to-patient teledermoscopy.^11,12

Machine Learning

Big data and machine learning has been hailed as the future of medicine and dermatology alike.¹³ Machine learning is a type of artificial intelligence that uses computational algorithms (eg, neural networks) that allow computer programs to automatically improve their accuracy (learn) by analyzing large data sets. In dermatology, machine learning has been most notably used to train computers to identify images of skin cancer by way of large image databases.^14-17 One algorithm, a convolutional neural network (CNN), made headlines in 2017 when it was able to identify dermoscopic and clinical images of skin cancer with comparable accuracy to a group of 21 dermatologists.¹⁴ In 2018, the International Skin Imaging Collaboration (ISIC) published results of a study of the diagnostic accuracy of 25 computer algorithms compared to 8 dermatologists using a set of 100 dermoscopic images of melanoma and benign nevi.¹⁵ Using the average sensitivity of the dermatologists (82%), the top fusion algorithm in the study had a sensitivity of 76% versus 59% for the dermatologists (P=.02). These results compared the mean sensitivity of the dermatologists, as some individual dermatologists outperformed the algorithm.¹⁵ More recently, another CNN was compared to 58 international dermatologists in the classification of a set of 100 dermoscopic images (20 melanoma and 80 melanocytic nevi).¹⁶ Using the mean sensitivity of the dermatologists (86.6%), the CNN had a specificity of 92.5% versus 71.3% for dermatologists (P<.01). In the second part of the study, the dermatologists were given some clinical information and close-up photographs of the lesions, which improved their average (SD) sensitivity and specificity to 88.9% (9.6%)(P=.19) and 75.7% (11.7%)(P<.05), respectively. When compared to the CNN at this higher sensitivity, the CNN still had a higher specificity than the dermatologists (82.5% vs 75.7% [P<.01]).¹⁶ However, in real-life clinical practice dermatologists perform better, not only because they can collect more in-person clinical information but also because humans gather more information during live examination than when they are interpreting close-up clinical and/or dermoscopic images. In a sense, we currently are limited to comparing data that is incommensurable.

Machine learning studies have other notable limitations, such as data sets that do not contain a full spectrum of skin lesions or less common lesions (eg, pigmented seborrheic keratoses, amelanotic melanomas) and variation in image databases used.^15,16 For machine algorithms to improve, they require access to high-quality and ideally standardized digital dermoscopic image databases. The ISIC and other organizations currently have databases specifically for this purpose, but more images are needed.¹⁸ As additional practitioners incorporate digital dermoscopy in their clinical practice, the potential for larger databases and more accurate algorithms becomes a possibility. 

Image Acquisition

Many devices are available for digital dermoscopic image acquisition, including dermatoscopes that attach to smartphones and/or digital cameras and all-in-one systems (eTable). The exact system employed will depend on the practitioner's requirements for price, portability, speed, image quality, and software. Digital single-lens reflex (DSLR) cameras boast the highest image quality, while video dermoscopy traditionally yields stored images with poor resolution.¹⁹ Macroscopic images obtained by other imaging devices, including spectral imaging devices and reflectance confocal microscopy, usually are yielded via video dermoscopy or a video camera to capture images; thus, stored images generally are not as high quality. 

Smartphones are increasingly used for clinical imaging in dermatology.²⁰ Although DSLR cameras still take the highest-quality images, current smartphone image quality is comparable to digital cameras.^21,22 Computational photography uses computer processing power to enhance image quality and may bring smartphone image quality closer to DSLR cameras.^22,23 Smartphones with newer dual-lens cameras have been reported to further improve image quality.²¹ Current smartphones have the option of enabling high-dynamic-range imaging, which combines multiple images taken with different exposures to create a single image with improved dynamic range of luminosity. It has been reported that high-dynamic-range imaging may even enhance dermoscopic features of more challenging hypopigmented skin cancers.²⁴

Standardizing Imaging

There has been a concerted effort to standardize digital dermatologic image acquisition.^25,26 Standardization promises to facilitate data analysis, improve collaboration, protect patient privacy, and improve patient care.^13,26,27 At the forefront of image standardization is the ISIC organization, which recently published its Delphi consensus guidelines on standards for lesion imaging, including dermoscopy.²⁶

The true holy grail of image standardization is the Digital Imaging and Communications in Medicine (DICOM) standard.^26-28 The DICOM is a comprehensive imaging standard for storage, annotation, transfer, and display of images, and it is most notable for its use in radiology. The DICOM also could be applied to new imaging modalities in dermatology (eg, optical coherence tomography, reflectance confocal microscopy). Past efforts to develop a DICOM standard for dermatology were undertaken by a working group that has since disbanded.²⁷ Work by the ISIC and many others will hopefully lead to adoption of the DICOM standard by dermatology at some point in the future. 

Protected Health Information

The Health Insurance Portability and Accountability Act (HIPAA) requires protected health information (PHI) to be stored in a secure manner with limited access that sufficiently protects identifiable patient information. Although dermoscopic images generally are deidentified, they often are stored alongside clinical photographs and data that contains PHI in clinical practice.

Image storage can take 2 forms: (1) physical local storage on internal and external hard drives or (2) remote storage (eg, cloud-based storage). Encryption is essential regardless of the method of storage. It is required by law that loss of nonencrypted PHI be reported to all potentially affected patients, the US Department of Health & Human Services, and local/state media depending on the number of patients affected. Loss of PHI can result in fines of up to $1.5 million.²⁹ On the contrary, loss of properly encrypted data would not be required to be reported.³⁰

As smart image acquisition devices begin to dominate the clinical setting, practitioners need to be vigilant in securing patient PHI. There are multiple applications (apps) that allow for secure encrypted digital dermoscopic image acquisition and storage on smartphones. Additionally, it is important to secure smartphones with complex passcodes (eg, a mix of special characters, numbers, uppercase and lowercase letters). Most dermatoscope manufacturers have apps for image acquisition and storage that can be tied into other platforms or storage systems (eg, DermLite app [3Gen], Handyscope [FotoFinder Systems GmbH], VEOS app [Canfield Scientific, Inc]).²⁸ Other options include syncing images with current electronic medical record technologies, transferring photographs to HIPAA-compliant cloud storage, or transferring photographs to an encrypted computer and/or external hard drive. Some tips for securing data based on HIPAA and other guidelines are listed in the Table.^30,31

Conclusion

The expansion of teledermoscopy alongside direct-to-patient services may create additional incentives for clinicians to incorporate digital dermoscopy into their practice. As more practitioners adopt digital dermoscopy, machine learning driven by technological advancements and larger image data sets could influence the future practice of dermatology. With the rise in digital dermoscopy by way of smartphones, additional steps must be taken to ensure patients' PHI is safeguarded. Digital dermoscopy is a dynamic field that will likely see continued growth in the coming years.

Dermoscopic examination has been proven to increase diagnostic accuracy and decrease unnecessary biopsies of both melanoma and nonmelanoma skin cancers.^1,2 Digital dermoscopy refers to acquiring and storing digital dermoscopic photographs via digital camera, smart image capture devices such as smartphones and tablets, or any other devices used for image acquisition. The stored images may then be used in a variety of ways, including sequential digital monitoring, teledermoscopy, and machine learning.

Sequential Digital Monitoring

Sequential digital dermoscopy imaging (SDDI) is the capture and storage of dermoscopic images of suspicious lesions that are then monitored over time for changes. Studies have shown that SDDI allows for early detection of melanomas and leads to a decrease in the number of unnecessary excisions.^3,4 A meta-analysis of SDDI found that the chance of detecting melanoma increased with the length of monitoring, which suggests that continued follow-up, especially in high-risk groups, is crucial.⁴

Teledermoscopy

Teledermatology (telederm) is on the rise in the United States, with the number of programs and consultations increasing yearly. One study showed a 48% increase in telederm programs in the last 5 years.⁵ Studies have shown the addition of digital dermoscopic images improved the diagnostic accuracy in telederm skin cancer screenings versus clinical images alone.^6,7

Telederm currently is practiced in 2 main models: live-interactive video consultation and storage of images for future consultation (store and forward). Medicare currently only reimburses live-interactive telederm for patients in nonmetropolitan areas and store-and-forward telederm pilot programs in Alaska and Hawaii; however, Medicaid does reimburse for store and forward in a handful of states.8 Similar to dermatoscope use during clinical examination, there currently is no additional reimbursement for teledermoscopy. Of note, a willingness-to-pay survey of 214 students from a southwestern university health center showed that participants were willing to pay an average (SD) of $55.27 ($39.11) out of pocket for a teledermoscopy/telederm evaluation, citing factors such as convenience.⁹

Direct-to-consumer telederm offers a new way for patients to receive care.¹⁰ Some dermatoscopes (eg, DermLite HÜD [3Gen], Molescope/Molescope II [Metaoptima Technology Inc]) currently are marketed directly to consumers along with telederm services to facilitate direct-to-patient teledermoscopy.^11,12

Machine Learning

Big data and machine learning has been hailed as the future of medicine and dermatology alike.¹³ Machine learning is a type of artificial intelligence that uses computational algorithms (eg, neural networks) that allow computer programs to automatically improve their accuracy (learn) by analyzing large data sets. In dermatology, machine learning has been most notably used to train computers to identify images of skin cancer by way of large image databases.^14-17 One algorithm, a convolutional neural network (CNN), made headlines in 2017 when it was able to identify dermoscopic and clinical images of skin cancer with comparable accuracy to a group of 21 dermatologists.¹⁴ In 2018, the International Skin Imaging Collaboration (ISIC) published results of a study of the diagnostic accuracy of 25 computer algorithms compared to 8 dermatologists using a set of 100 dermoscopic images of melanoma and benign nevi.¹⁵ Using the average sensitivity of the dermatologists (82%), the top fusion algorithm in the study had a sensitivity of 76% versus 59% for the dermatologists (P=.02). These results compared the mean sensitivity of the dermatologists, as some individual dermatologists outperformed the algorithm.¹⁵ More recently, another CNN was compared to 58 international dermatologists in the classification of a set of 100 dermoscopic images (20 melanoma and 80 melanocytic nevi).¹⁶ Using the mean sensitivity of the dermatologists (86.6%), the CNN had a specificity of 92.5% versus 71.3% for dermatologists (P<.01). In the second part of the study, the dermatologists were given some clinical information and close-up photographs of the lesions, which improved their average (SD) sensitivity and specificity to 88.9% (9.6%)(P=.19) and 75.7% (11.7%)(P<.05), respectively. When compared to the CNN at this higher sensitivity, the CNN still had a higher specificity than the dermatologists (82.5% vs 75.7% [P<.01]).¹⁶ However, in real-life clinical practice dermatologists perform better, not only because they can collect more in-person clinical information but also because humans gather more information during live examination than when they are interpreting close-up clinical and/or dermoscopic images. In a sense, we currently are limited to comparing data that is incommensurable.

Machine learning studies have other notable limitations, such as data sets that do not contain a full spectrum of skin lesions or less common lesions (eg, pigmented seborrheic keratoses, amelanotic melanomas) and variation in image databases used.^15,16 For machine algorithms to improve, they require access to high-quality and ideally standardized digital dermoscopic image databases. The ISIC and other organizations currently have databases specifically for this purpose, but more images are needed.¹⁸ As additional practitioners incorporate digital dermoscopy in their clinical practice, the potential for larger databases and more accurate algorithms becomes a possibility. 

Image Acquisition

Many devices are available for digital dermoscopic image acquisition, including dermatoscopes that attach to smartphones and/or digital cameras and all-in-one systems (eTable). The exact system employed will depend on the practitioner's requirements for price, portability, speed, image quality, and software. Digital single-lens reflex (DSLR) cameras boast the highest image quality, while video dermoscopy traditionally yields stored images with poor resolution.¹⁹ Macroscopic images obtained by other imaging devices, including spectral imaging devices and reflectance confocal microscopy, usually are yielded via video dermoscopy or a video camera to capture images; thus, stored images generally are not as high quality. 

Smartphones are increasingly used for clinical imaging in dermatology.²⁰ Although DSLR cameras still take the highest-quality images, current smartphone image quality is comparable to digital cameras.^21,22 Computational photography uses computer processing power to enhance image quality and may bring smartphone image quality closer to DSLR cameras.^22,23 Smartphones with newer dual-lens cameras have been reported to further improve image quality.²¹ Current smartphones have the option of enabling high-dynamic-range imaging, which combines multiple images taken with different exposures to create a single image with improved dynamic range of luminosity. It has been reported that high-dynamic-range imaging may even enhance dermoscopic features of more challenging hypopigmented skin cancers.²⁴

Standardizing Imaging

There has been a concerted effort to standardize digital dermatologic image acquisition.^25,26 Standardization promises to facilitate data analysis, improve collaboration, protect patient privacy, and improve patient care.^13,26,27 At the forefront of image standardization is the ISIC organization, which recently published its Delphi consensus guidelines on standards for lesion imaging, including dermoscopy.²⁶

The true holy grail of image standardization is the Digital Imaging and Communications in Medicine (DICOM) standard.^26-28 The DICOM is a comprehensive imaging standard for storage, annotation, transfer, and display of images, and it is most notable for its use in radiology. The DICOM also could be applied to new imaging modalities in dermatology (eg, optical coherence tomography, reflectance confocal microscopy). Past efforts to develop a DICOM standard for dermatology were undertaken by a working group that has since disbanded.²⁷ Work by the ISIC and many others will hopefully lead to adoption of the DICOM standard by dermatology at some point in the future. 

Protected Health Information

The Health Insurance Portability and Accountability Act (HIPAA) requires protected health information (PHI) to be stored in a secure manner with limited access that sufficiently protects identifiable patient information. Although dermoscopic images generally are deidentified, they often are stored alongside clinical photographs and data that contains PHI in clinical practice.

Image storage can take 2 forms: (1) physical local storage on internal and external hard drives or (2) remote storage (eg, cloud-based storage). Encryption is essential regardless of the method of storage. It is required by law that loss of nonencrypted PHI be reported to all potentially affected patients, the US Department of Health & Human Services, and local/state media depending on the number of patients affected. Loss of PHI can result in fines of up to $1.5 million.²⁹ On the contrary, loss of properly encrypted data would not be required to be reported.³⁰

As smart image acquisition devices begin to dominate the clinical setting, practitioners need to be vigilant in securing patient PHI. There are multiple applications (apps) that allow for secure encrypted digital dermoscopic image acquisition and storage on smartphones. Additionally, it is important to secure smartphones with complex passcodes (eg, a mix of special characters, numbers, uppercase and lowercase letters). Most dermatoscope manufacturers have apps for image acquisition and storage that can be tied into other platforms or storage systems (eg, DermLite app [3Gen], Handyscope [FotoFinder Systems GmbH], VEOS app [Canfield Scientific, Inc]).²⁸ Other options include syncing images with current electronic medical record technologies, transferring photographs to HIPAA-compliant cloud storage, or transferring photographs to an encrypted computer and/or external hard drive. Some tips for securing data based on HIPAA and other guidelines are listed in the Table.^30,31

Conclusion

The expansion of teledermoscopy alongside direct-to-patient services may create additional incentives for clinicians to incorporate digital dermoscopy into their practice. As more practitioners adopt digital dermoscopy, machine learning driven by technological advancements and larger image data sets could influence the future practice of dermatology. With the rise in digital dermoscopy by way of smartphones, additional steps must be taken to ensure patients' PHI is safeguarded. Digital dermoscopy is a dynamic field that will likely see continued growth in the coming years.

Skin cancer is diagnosed in approximately 5.4 million individuals annually in the United States, more than the total number of breast, lung, colon, and prostate cancers diagnosed per year.¹ It is estimated that 1 in 5 Americans will develop skin cancer during their lifetime.² The 2 most common forms of skin cancer are basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), accounting for 4 million and 1 million cases diagnosed each year, respectively.³ With the increasing incidence of these skin cancers, the use of noninvasive imaging tools for detection and diagnosis has grown.

Ex vivo confocal microscopy is a diagnostic imaging tool that can be used in real-time at the bedside to assess freshly excised tissue for malignancies. It images tissue samples with cellular resolution and within minutes of biopsy or excision. Ex vivo confocal microscopy is a versatile tool that can assist in the diagnosis and management of skin malignancies such as melanoma, BCC, and SCC.

Reflectance vs Fluorescence Mode

Excised lesions can be examined in reflectance or fluorescence mode in great detail but with slightly varying nuclear-to-dermis contrasts depending on the chromophore that is targeted. In reflectance mode (reflectance confocal microscopy [RCM]), melanin and keratin act as endogenous chromophores because of their high refractive index relative to water,^4,5 which allows for the visualization of cellular structures of the skin at low power, as well as microscopic substructures such as melanosomes, cytoplasmic granules, and other cellular organelles at high power. Although an exogenous contrast agent is not required, acetic acid has the capability to highlight nuclei, enhancing the tumor cell-to-dermis contrast in RCM.⁶ Acetic acid is clinically used as a predictor for certain skin and mucosal membrane neoplasms that blanch when exposed to the solution. In the case of RCM, acetic acid increases the visibility of nuclei by inducing the compaction of chromatin. For the acetowhitening to be effective, the sample must be soaked in the solution for a specific amount of time, depending on the concentration.⁷ A concentration between 1% and 10% can be used, but the less concentrated the solution, the longer the time of soaking that is required to achieve sufficiently bright nuclei.⁶

The contrast with acetic acid, however, is quite weak when the tissue is imaged en face, or along the horizontal surface of the sample, due to the collagen in the dermal layer, which has a high reflectance index. This issue is rectified when using the confocal microscope in the fluorescence mode with an exogenous fluorescent dye as a nuclear stain. Fluorescence confocal microscopy (FCM), results in a stronger nuclear-to-dermal contrast because of the role of contrast agents.⁸ The 1000-fold increase in contrast between nuclei and dermis is the result of dye agents that preferentially bind to nuclear DNA, of which acridine orange is the most commonly used.^5,8 Basal cell carcinoma and SCC tumor cells can be visualized with FCM because they appear hyperfluorescent when stained with acridine orange.⁹ The acridine orange–stained cells display bright nuclei, while the cytoplasm and collagen remains dark. A positive feature of acridine orange is that it does not alter the tissue sample during freezing or formalin fixation and thus has no effect on subsequent histopathology that may need to be performed on the sample.¹⁰

High-Resolution Images Aid in Diagnosis

After it is harvested, the tissue sample is soaked in a contrast agent or dye, if needed, depending on the confocal mode to be used. The confocal microscope is then used to take a series of high-resolution individual en face images that are then stitched together to create a final mosaic image that can be up to 12×12 mm.^6,11 With a 200-µm depth visibility, confocal microscopy can capture the cellular structures in the epidermis, dermis, and (if compressed enough) subcutaneous fat in just under 3 minutes.¹²

The images produced through confocal microscopy have an excellent correlation to frozen histological sections and can aid in the diagnosis of many epidermal and dermal malignancies including melanoma, BCC, and SCC. New criteria have been established to aid in the interpretation of the confocal images and identify some of the more common skin cancers.^5,12,13 Basal cell carcinoma samples imaged through fluorescence and reflectance in low-power mode display the distinct nodular patterns with well-demarcated edges, as seen on classical histopathology. In the case of FCM, the cells that make up the tumor display hyperfluorescent areas consistent with nucleated cells that are stained with acridine orange. The main features that identify BCC on FCM images include nuclear pleomorphism and crowding, peripheral palisading, clefting of the basaloid islands, increased nucleus-to-cytoplasm ratio, and the presence of a modified dermis surrounding the mass known as the tumoral stroma^5,12 (Figure).

In addition to fluorescence and a well-defined tumor silhouette, SCC under FCM displays keratin pearls composed of keratinized squames, nuclear pleomorphism, and fluorescent scales in the stratum corneum that are a result of keratin formation.^5,13 The extent of differentiation of the SCC lesion also can be determined by assessing if the silhouette is well defined. A well-defined tumor silhouette is consistent with the diagnosis of a well-differentiated SCC, and vice versa.¹³ Ex vivo RCM also has been shown to be useful in diagnosing malignant melanomas, with melanin acting as an endogenous chromophore. Some of the features seen on imaging include a disarranged epithelium, hyperreflective roundish and dendritic pagetoid cells, and large hyperreflective polymorphic cells in the superficial chorion.¹⁴

Comparison to Conventional Histopathology

Ex vivo confocal microscopy in both the reflectance and fluorescence mode has been shown to perform well compared to conventional histopathology in the diagnosis of biopsy specimens. Ex vivo FCM has been shown to have an overall sensitivity of 88% and specificity of 99% in detecting residual BCC at the margins of excised tissue samples and in the fraction of the time it takes to attain similar results with frozen histopathology.⁹ Ex vivo RCM has been shown to have a higher prognostic capability, with 100% sensitivity and specificity in identifying BCC when scanning the tissue samples en face.¹⁵

Qualitatively, the images produced by RCM and FCM are similar to histopathology in overall architecture. Both techniques enhance the contrast between the epithelium and stroma and create images that can be examined in low as well as high resolution. A substantial difference between confocal microscopy and conventional hematoxylin and eosin–stained histopathology is that the confocal microscope produces images in gray scale. One way to alter the black-and-white images to resemble hematoxylin and eosin–stained slides is through the use of digital staining,¹⁶ which could boost clinical acceptance by physicians who are accustomed to the classical pink-purple appearance of pathology slides and could potentially limit the learning curve needed to read the confocal images.

Application in Mohs Micrographic Surgery

An important clinical application of ex vivo FCM imaging that has emerged is the detection of malignant cells at the excision margins during Mohs micrographic surgery. The use of confocal microscopy has the potential to save time by eliminating the need for tissue fixation while still providing good diagnostic accuracy. Implementing FCM as an imaging tool to guide surgical excisions could provide rapid diagnosis of the tissue, expediting excisions and reconstruction or the Mohs procedure while eliminating patient wait time and the need for frozen histopathology. Ex vivo RCM also has been used to establish laser parameters for CO₂ laser ablation of superficial and early nodular BCC lesions.¹⁷ Other potential uses for ex vivo RCM/FCM could include rapid evaluation of tissue during operating room procedures where rapid frozen sections are currently utilized.

Combining In Vivo and Ex Vivo Confocal Microscopy

Many of the diagnostic guidelines created with the use of ex vivo confocal microscopy have been applied to in vivo use, and therefore the use of both modalities is appealing. In vivo confocal microscopy is a noninvasive technique that has been used to map margins of skin tumors such as BCC and lentigo maligna at the bedside.⁵ It also has been shown to help plan both surgical and nonsurgical treatment modalities and reconstruction before the tumor is excised.¹⁸ This technique also can help the patient understand the extent of the excision and any subsequent reconstruction that may be needed.

Limitations

Ex vivo confocal microscopy used as a diagnostic tool does have some limitations. Its novelty may require surgeons and pathologists to be trained to interpret the images properly and correlate them to conventional diagnostic guidelines. The imaging also is limited to a depth of approximately 200 µm; however, the sample may be flipped so that the underside can be imaged as well, which increases the depth to approximately 400 µm. The tissue being imaged must be fixed flat, which may alter its shape. Complex tissue samples may be difficult to flatten out completely and therefore may be difficult to image. A special mount may be required for the sample to be fixed in a proper position for imaging.⁶

Final Thoughts

Despite some of these limitations, the need for rapid bedside tissue diagnosis makes ex vivo confocal microscopy an attractive device that can be used as an additional diagnostic tool to histopathology and also has been tested in other disciplines, such as breast cancer pathology. In the future, both in vivo and ex vivo confocal microscopy may be utilized to diagnose cutaneous malignancies, guide surgical excisions, and detect lesion progression, and it may become a basis for rapid diagnosis and detection.¹⁹

Skin cancer is diagnosed in approximately 5.4 million individuals annually in the United States, more than the total number of breast, lung, colon, and prostate cancers diagnosed per year.¹ It is estimated that 1 in 5 Americans will develop skin cancer during their lifetime.² The 2 most common forms of skin cancer are basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), accounting for 4 million and 1 million cases diagnosed each year, respectively.³ With the increasing incidence of these skin cancers, the use of noninvasive imaging tools for detection and diagnosis has grown.

Ex vivo confocal microscopy is a diagnostic imaging tool that can be used in real-time at the bedside to assess freshly excised tissue for malignancies. It images tissue samples with cellular resolution and within minutes of biopsy or excision. Ex vivo confocal microscopy is a versatile tool that can assist in the diagnosis and management of skin malignancies such as melanoma, BCC, and SCC.

Reflectance vs Fluorescence Mode

Excised lesions can be examined in reflectance or fluorescence mode in great detail but with slightly varying nuclear-to-dermis contrasts depending on the chromophore that is targeted. In reflectance mode (reflectance confocal microscopy [RCM]), melanin and keratin act as endogenous chromophores because of their high refractive index relative to water,^4,5 which allows for the visualization of cellular structures of the skin at low power, as well as microscopic substructures such as melanosomes, cytoplasmic granules, and other cellular organelles at high power. Although an exogenous contrast agent is not required, acetic acid has the capability to highlight nuclei, enhancing the tumor cell-to-dermis contrast in RCM.⁶ Acetic acid is clinically used as a predictor for certain skin and mucosal membrane neoplasms that blanch when exposed to the solution. In the case of RCM, acetic acid increases the visibility of nuclei by inducing the compaction of chromatin. For the acetowhitening to be effective, the sample must be soaked in the solution for a specific amount of time, depending on the concentration.⁷ A concentration between 1% and 10% can be used, but the less concentrated the solution, the longer the time of soaking that is required to achieve sufficiently bright nuclei.⁶

The contrast with acetic acid, however, is quite weak when the tissue is imaged en face, or along the horizontal surface of the sample, due to the collagen in the dermal layer, which has a high reflectance index. This issue is rectified when using the confocal microscope in the fluorescence mode with an exogenous fluorescent dye as a nuclear stain. Fluorescence confocal microscopy (FCM), results in a stronger nuclear-to-dermal contrast because of the role of contrast agents.⁸ The 1000-fold increase in contrast between nuclei and dermis is the result of dye agents that preferentially bind to nuclear DNA, of which acridine orange is the most commonly used.^5,8 Basal cell carcinoma and SCC tumor cells can be visualized with FCM because they appear hyperfluorescent when stained with acridine orange.⁹ The acridine orange–stained cells display bright nuclei, while the cytoplasm and collagen remains dark. A positive feature of acridine orange is that it does not alter the tissue sample during freezing or formalin fixation and thus has no effect on subsequent histopathology that may need to be performed on the sample.¹⁰

High-Resolution Images Aid in Diagnosis

After it is harvested, the tissue sample is soaked in a contrast agent or dye, if needed, depending on the confocal mode to be used. The confocal microscope is then used to take a series of high-resolution individual en face images that are then stitched together to create a final mosaic image that can be up to 12×12 mm.^6,11 With a 200-µm depth visibility, confocal microscopy can capture the cellular structures in the epidermis, dermis, and (if compressed enough) subcutaneous fat in just under 3 minutes.¹²

The images produced through confocal microscopy have an excellent correlation to frozen histological sections and can aid in the diagnosis of many epidermal and dermal malignancies including melanoma, BCC, and SCC. New criteria have been established to aid in the interpretation of the confocal images and identify some of the more common skin cancers.^5,12,13 Basal cell carcinoma samples imaged through fluorescence and reflectance in low-power mode display the distinct nodular patterns with well-demarcated edges, as seen on classical histopathology. In the case of FCM, the cells that make up the tumor display hyperfluorescent areas consistent with nucleated cells that are stained with acridine orange. The main features that identify BCC on FCM images include nuclear pleomorphism and crowding, peripheral palisading, clefting of the basaloid islands, increased nucleus-to-cytoplasm ratio, and the presence of a modified dermis surrounding the mass known as the tumoral stroma^5,12 (Figure).

In addition to fluorescence and a well-defined tumor silhouette, SCC under FCM displays keratin pearls composed of keratinized squames, nuclear pleomorphism, and fluorescent scales in the stratum corneum that are a result of keratin formation.^5,13 The extent of differentiation of the SCC lesion also can be determined by assessing if the silhouette is well defined. A well-defined tumor silhouette is consistent with the diagnosis of a well-differentiated SCC, and vice versa.¹³ Ex vivo RCM also has been shown to be useful in diagnosing malignant melanomas, with melanin acting as an endogenous chromophore. Some of the features seen on imaging include a disarranged epithelium, hyperreflective roundish and dendritic pagetoid cells, and large hyperreflective polymorphic cells in the superficial chorion.¹⁴

Comparison to Conventional Histopathology

Ex vivo confocal microscopy in both the reflectance and fluorescence mode has been shown to perform well compared to conventional histopathology in the diagnosis of biopsy specimens. Ex vivo FCM has been shown to have an overall sensitivity of 88% and specificity of 99% in detecting residual BCC at the margins of excised tissue samples and in the fraction of the time it takes to attain similar results with frozen histopathology.⁹ Ex vivo RCM has been shown to have a higher prognostic capability, with 100% sensitivity and specificity in identifying BCC when scanning the tissue samples en face.¹⁵

Qualitatively, the images produced by RCM and FCM are similar to histopathology in overall architecture. Both techniques enhance the contrast between the epithelium and stroma and create images that can be examined in low as well as high resolution. A substantial difference between confocal microscopy and conventional hematoxylin and eosin–stained histopathology is that the confocal microscope produces images in gray scale. One way to alter the black-and-white images to resemble hematoxylin and eosin–stained slides is through the use of digital staining,¹⁶ which could boost clinical acceptance by physicians who are accustomed to the classical pink-purple appearance of pathology slides and could potentially limit the learning curve needed to read the confocal images.

Application in Mohs Micrographic Surgery

An important clinical application of ex vivo FCM imaging that has emerged is the detection of malignant cells at the excision margins during Mohs micrographic surgery. The use of confocal microscopy has the potential to save time by eliminating the need for tissue fixation while still providing good diagnostic accuracy. Implementing FCM as an imaging tool to guide surgical excisions could provide rapid diagnosis of the tissue, expediting excisions and reconstruction or the Mohs procedure while eliminating patient wait time and the need for frozen histopathology. Ex vivo RCM also has been used to establish laser parameters for CO₂ laser ablation of superficial and early nodular BCC lesions.¹⁷ Other potential uses for ex vivo RCM/FCM could include rapid evaluation of tissue during operating room procedures where rapid frozen sections are currently utilized.

Combining In Vivo and Ex Vivo Confocal Microscopy

Many of the diagnostic guidelines created with the use of ex vivo confocal microscopy have been applied to in vivo use, and therefore the use of both modalities is appealing. In vivo confocal microscopy is a noninvasive technique that has been used to map margins of skin tumors such as BCC and lentigo maligna at the bedside.⁵ It also has been shown to help plan both surgical and nonsurgical treatment modalities and reconstruction before the tumor is excised.¹⁸ This technique also can help the patient understand the extent of the excision and any subsequent reconstruction that may be needed.

Limitations

Ex vivo confocal microscopy used as a diagnostic tool does have some limitations. Its novelty may require surgeons and pathologists to be trained to interpret the images properly and correlate them to conventional diagnostic guidelines. The imaging also is limited to a depth of approximately 200 µm; however, the sample may be flipped so that the underside can be imaged as well, which increases the depth to approximately 400 µm. The tissue being imaged must be fixed flat, which may alter its shape. Complex tissue samples may be difficult to flatten out completely and therefore may be difficult to image. A special mount may be required for the sample to be fixed in a proper position for imaging.⁶

Final Thoughts

Despite some of these limitations, the need for rapid bedside tissue diagnosis makes ex vivo confocal microscopy an attractive device that can be used as an additional diagnostic tool to histopathology and also has been tested in other disciplines, such as breast cancer pathology. In the future, both in vivo and ex vivo confocal microscopy may be utilized to diagnose cutaneous malignancies, guide surgical excisions, and detect lesion progression, and it may become a basis for rapid diagnosis and detection.¹⁹

Skin cancer is diagnosed in approximately 5.4 million individuals annually in the United States, more than the total number of breast, lung, colon, and prostate cancers diagnosed per year.¹ It is estimated that 1 in 5 Americans will develop skin cancer during their lifetime.² The 2 most common forms of skin cancer are basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), accounting for 4 million and 1 million cases diagnosed each year, respectively.³ With the increasing incidence of these skin cancers, the use of noninvasive imaging tools for detection and diagnosis has grown.

Ex vivo confocal microscopy is a diagnostic imaging tool that can be used in real-time at the bedside to assess freshly excised tissue for malignancies. It images tissue samples with cellular resolution and within minutes of biopsy or excision. Ex vivo confocal microscopy is a versatile tool that can assist in the diagnosis and management of skin malignancies such as melanoma, BCC, and SCC.

Reflectance vs Fluorescence Mode

Excised lesions can be examined in reflectance or fluorescence mode in great detail but with slightly varying nuclear-to-dermis contrasts depending on the chromophore that is targeted. In reflectance mode (reflectance confocal microscopy [RCM]), melanin and keratin act as endogenous chromophores because of their high refractive index relative to water,^4,5 which allows for the visualization of cellular structures of the skin at low power, as well as microscopic substructures such as melanosomes, cytoplasmic granules, and other cellular organelles at high power. Although an exogenous contrast agent is not required, acetic acid has the capability to highlight nuclei, enhancing the tumor cell-to-dermis contrast in RCM.⁶ Acetic acid is clinically used as a predictor for certain skin and mucosal membrane neoplasms that blanch when exposed to the solution. In the case of RCM, acetic acid increases the visibility of nuclei by inducing the compaction of chromatin. For the acetowhitening to be effective, the sample must be soaked in the solution for a specific amount of time, depending on the concentration.⁷ A concentration between 1% and 10% can be used, but the less concentrated the solution, the longer the time of soaking that is required to achieve sufficiently bright nuclei.⁶

The contrast with acetic acid, however, is quite weak when the tissue is imaged en face, or along the horizontal surface of the sample, due to the collagen in the dermal layer, which has a high reflectance index. This issue is rectified when using the confocal microscope in the fluorescence mode with an exogenous fluorescent dye as a nuclear stain. Fluorescence confocal microscopy (FCM), results in a stronger nuclear-to-dermal contrast because of the role of contrast agents.⁸ The 1000-fold increase in contrast between nuclei and dermis is the result of dye agents that preferentially bind to nuclear DNA, of which acridine orange is the most commonly used.^5,8 Basal cell carcinoma and SCC tumor cells can be visualized with FCM because they appear hyperfluorescent when stained with acridine orange.⁹ The acridine orange–stained cells display bright nuclei, while the cytoplasm and collagen remains dark. A positive feature of acridine orange is that it does not alter the tissue sample during freezing or formalin fixation and thus has no effect on subsequent histopathology that may need to be performed on the sample.¹⁰

High-Resolution Images Aid in Diagnosis

After it is harvested, the tissue sample is soaked in a contrast agent or dye, if needed, depending on the confocal mode to be used. The confocal microscope is then used to take a series of high-resolution individual en face images that are then stitched together to create a final mosaic image that can be up to 12×12 mm.^6,11 With a 200-µm depth visibility, confocal microscopy can capture the cellular structures in the epidermis, dermis, and (if compressed enough) subcutaneous fat in just under 3 minutes.¹²

The images produced through confocal microscopy have an excellent correlation to frozen histological sections and can aid in the diagnosis of many epidermal and dermal malignancies including melanoma, BCC, and SCC. New criteria have been established to aid in the interpretation of the confocal images and identify some of the more common skin cancers.^5,12,13 Basal cell carcinoma samples imaged through fluorescence and reflectance in low-power mode display the distinct nodular patterns with well-demarcated edges, as seen on classical histopathology. In the case of FCM, the cells that make up the tumor display hyperfluorescent areas consistent with nucleated cells that are stained with acridine orange. The main features that identify BCC on FCM images include nuclear pleomorphism and crowding, peripheral palisading, clefting of the basaloid islands, increased nucleus-to-cytoplasm ratio, and the presence of a modified dermis surrounding the mass known as the tumoral stroma^5,12 (Figure).

In addition to fluorescence and a well-defined tumor silhouette, SCC under FCM displays keratin pearls composed of keratinized squames, nuclear pleomorphism, and fluorescent scales in the stratum corneum that are a result of keratin formation.^5,13 The extent of differentiation of the SCC lesion also can be determined by assessing if the silhouette is well defined. A well-defined tumor silhouette is consistent with the diagnosis of a well-differentiated SCC, and vice versa.¹³ Ex vivo RCM also has been shown to be useful in diagnosing malignant melanomas, with melanin acting as an endogenous chromophore. Some of the features seen on imaging include a disarranged epithelium, hyperreflective roundish and dendritic pagetoid cells, and large hyperreflective polymorphic cells in the superficial chorion.¹⁴

Comparison to Conventional Histopathology

Ex vivo confocal microscopy in both the reflectance and fluorescence mode has been shown to perform well compared to conventional histopathology in the diagnosis of biopsy specimens. Ex vivo FCM has been shown to have an overall sensitivity of 88% and specificity of 99% in detecting residual BCC at the margins of excised tissue samples and in the fraction of the time it takes to attain similar results with frozen histopathology.⁹ Ex vivo RCM has been shown to have a higher prognostic capability, with 100% sensitivity and specificity in identifying BCC when scanning the tissue samples en face.¹⁵

Qualitatively, the images produced by RCM and FCM are similar to histopathology in overall architecture. Both techniques enhance the contrast between the epithelium and stroma and create images that can be examined in low as well as high resolution. A substantial difference between confocal microscopy and conventional hematoxylin and eosin–stained histopathology is that the confocal microscope produces images in gray scale. One way to alter the black-and-white images to resemble hematoxylin and eosin–stained slides is through the use of digital staining,¹⁶ which could boost clinical acceptance by physicians who are accustomed to the classical pink-purple appearance of pathology slides and could potentially limit the learning curve needed to read the confocal images.

Application in Mohs Micrographic Surgery

An important clinical application of ex vivo FCM imaging that has emerged is the detection of malignant cells at the excision margins during Mohs micrographic surgery. The use of confocal microscopy has the potential to save time by eliminating the need for tissue fixation while still providing good diagnostic accuracy. Implementing FCM as an imaging tool to guide surgical excisions could provide rapid diagnosis of the tissue, expediting excisions and reconstruction or the Mohs procedure while eliminating patient wait time and the need for frozen histopathology. Ex vivo RCM also has been used to establish laser parameters for CO₂ laser ablation of superficial and early nodular BCC lesions.¹⁷ Other potential uses for ex vivo RCM/FCM could include rapid evaluation of tissue during operating room procedures where rapid frozen sections are currently utilized.

Combining In Vivo and Ex Vivo Confocal Microscopy

Many of the diagnostic guidelines created with the use of ex vivo confocal microscopy have been applied to in vivo use, and therefore the use of both modalities is appealing. In vivo confocal microscopy is a noninvasive technique that has been used to map margins of skin tumors such as BCC and lentigo maligna at the bedside.⁵ It also has been shown to help plan both surgical and nonsurgical treatment modalities and reconstruction before the tumor is excised.¹⁸ This technique also can help the patient understand the extent of the excision and any subsequent reconstruction that may be needed.

Limitations

Ex vivo confocal microscopy used as a diagnostic tool does have some limitations. Its novelty may require surgeons and pathologists to be trained to interpret the images properly and correlate them to conventional diagnostic guidelines. The imaging also is limited to a depth of approximately 200 µm; however, the sample may be flipped so that the underside can be imaged as well, which increases the depth to approximately 400 µm. The tissue being imaged must be fixed flat, which may alter its shape. Complex tissue samples may be difficult to flatten out completely and therefore may be difficult to image. A special mount may be required for the sample to be fixed in a proper position for imaging.⁶

Final Thoughts

Despite some of these limitations, the need for rapid bedside tissue diagnosis makes ex vivo confocal microscopy an attractive device that can be used as an additional diagnostic tool to histopathology and also has been tested in other disciplines, such as breast cancer pathology. In the future, both in vivo and ex vivo confocal microscopy may be utilized to diagnose cutaneous malignancies, guide surgical excisions, and detect lesion progression, and it may become a basis for rapid diagnosis and detection.¹⁹

Reflectance confocal microscopy (RCM) imaging received Category I Current Procedural Terminology (CPT) codes by the Centers for Medicare & Medicaid Services in January 2016 and can now be submitted to insurance companies with reimbursement comparable to a skin biopsy or a global skin pathology service.¹ This fairly new technology is a US Food and Drug Administration–cleared noninvasive imaging modality that provides high-resolution in vivo cellular images of the skin. It has been shown to be efficacious in differentiating benign and malignant skin lesions, increasing diagnostic accuracy, and reducing the number of unnecessary skin biopsies that are performed. In addition to skin cancer diagnosis, RCM imaging also can help guide management of malignant lesions by detecting lateral margins prior to surgery as well as monitoring the lesion over time for treatment efficacy or recurrence. The potential impact of RCM imaging is tremendous, and reimbursement may lead to increased use in clinical practice to the benefit of our patients. Herein, we present a brief review of RCM imaging and reimbursement as well as the benefits and limitations of this new technology for dermatologists.

Reflectance Confocal Microscopy

In vivo RCM allows us to visualize the epidermis in real time on a cellular level down to the papillary dermis at a high resolution (×30) comparable to histologic examination. With optical sections 3- to 5-µm thick and a lateral resolution of 0.5 to 1.0 µm, RCM produces a stack of 500×500-µm² images up to a depth of approximately 200 µm.^2,3 At any chosen depth, these smaller images are stitched together with sophisticated software into a block, or mosaic, increasing the field of view to up to 8×8 mm². Imaging is performed in en face planes oriented parallel to the skin surface, similar to dermoscopy.

Current CPT Guidelines and Reimbursement

The CPT codes for RCM imaging provide reimbursement on a per-lesion basis and are similar to those used for skin biopsy and pathology (Table).¹ Codes 96931 through 96933 are used for imaging of a single lesion on a patient. The first code—96931—is used when image acquisition, interpretation, and report creation are carried out by a single clinician. The next 2 codes are used when one clinician acquires the image—96932—comparable to the technical component of a pathology code, while another reads it and creates the report—96933—similar to a dermatopathologist billing for the professional component of a pathology report. For patients presenting with multiple lesions, the next 3 codes—96934, 96935, and 96936—are used in conjunction with the applicable first code for each additional lesion with similar global, technical, and professional components. Because these codes are not in the radiology or pathology sections of CPT, a single code cannot be used with modifier -TC (technical component) and modifier -26, as they are in those sections.

The wide-probe VivaScope 1500 (Caliber I.D., Inc) currently is the only confocal device that can be reported with a CPT code and routinely reimbursed. The handheld VivaScope 3000 (Caliber I.D., Inc) can only view a small stack and does not have the ability to acquire a full mosaic image; it is not covered by these codes.

Images can be viewed as a stack captured at the same horizontal position but at sequential depths or as a mosaic, which has a larger field of view but is limited to a single plane. To appropriately assess a lesion, clinicians must obtain a mosaic that needs to be assessed at multiple layers for a diagnosis to be made because it is a cross-section view.

Diagnosis

Studies have demonstrated the usefulness of RCM imaging in the diagnosis of a wide range of skin diseases, including melanoma and nonmelanoma skin cancers, infectious diseases, and inflammatory and autoimmune conditions, as well as wound healing and skin aging. Reflectance confocal microscopy imaging is not limited to the skin; it can be used to evaluate the hair, nails, oral mucosa, and other organs.

According to several studies, RCM imaging notably increases the diagnostic accuracy and detection rate of skin cancers over clinical and dermoscopic examination alone and therefore can act as an aid in differentiating lesions that are benign versus those that are suspicious and should be biopsied.

Reflectance confocal microscopy has been shown to have a mean sensitivity of 94% (range, 92%–96%) and specificity of 83% (range, 81%–84%) for all types of skin cancer when used with dermoscopy.⁴ In particular, for melanocytic lesions that are ambiguous on dermoscopy, RCM used in addition to dermoscopy increases the mean sensitivity and specificity for melanoma diagnosis to 93% (range, 89%–96%) and 76% (range, 68%–83%), respectively.⁵ Although these reported sensitivities are comparable to dermoscopy, the specificity is superior, especially for detecting hypomelanotic and amelanotic melanomas, which often lack specific features on dermoscopy.^6-8

The combination of RCM with dermoscopy has reduced the number of unnecessary excisions of benign nevi by more than 50% when compared to dermoscopy alone.⁹ One study showed that the number needed to treat (ie, excise) a melanoma decreased from 14.6 with dermoscopy alone to 6.8 when guided by dermoscopy and RCM imaging.⁹ In a similar study, the number needed to treat dropped from 19.41 with dermoscopy alone to 6.25 with dermoscopy and RCM.¹⁰

These studies were not looking to evaluate RCM as a replacement test but rather as an add-on test to dermoscopy. Reflectance confocal microscopy imaging takes longer than dermoscopy for each lesion; therefore, RCM should only be used as an adjunctive tool to dermoscopy and not as an initial screening test. Consequentially, a dermatologist skilled in dermoscopy is essential in deciding which lesions would be appropriate for subsequent RCM imaging.

In Vivo Margin Mapping as an Adjunct to Surgery

Oftentimes, tumor margins are poorly defined and can be difficult to map clinically and dermoscopically. Studies have demonstrated the use of RCM in delineation of surgical margins prior to surgery or excisional biopsies.^11,12 Alternatively, when complete removal at biopsy would be impractical (eg, for extremely large lesions or lesions located in cosmetically sensitive areas such as the face), RCM can be used to pick the best site for an appropriate biopsy, which decreases the chance of sampling error due to skip lesions and increases histologic accuracy.

Nonsurgical Treatment Monitoring

One advantage of RCM over conventional histology is that RCM imaging leaves the tissue intact, allowing dynamic changes to be studied over time, which is useful for monitoring nonmelanoma skin cancers and lentigo maligna being treated with noninvasive therapeutic modalities.¹³ If not as a definitive treatment, RCM can act as an adjunct for surgery by monitoring reduction in lesion size prior to Mohs micrographic surgery, thereby decreasing the resulting surgical defect.¹⁴

Limitations

Imaging Depth
Although RCM is a revolutionary device in the field of dermatology, it has several limitations. With a maximal imaging depth of 350 µm, the imaging resolution decreases substantially with depth, limiting accurate interpretation to 200 µm. Reflectance confocal microscopy can only image the superficial portion of a lesion; therefore, deep tumor margins cannot be assessed. Hypertrophic or hyperkeratotic lesions, including lesions on the palms and soles, also are unable to be imaged with RCM. This limitation in depth penetration makes treatment monitoring impossible for invasive lesions that extend into the dermal layer.

Difficult-to-Reach Areas
Another limitation is the difficulty imaging areas such as the ocular canthi, nasal alae, or helices of the ear due to the wide probe size on the VivaScope 1500. The advent of the smaller handheld VivaScope 3000 device allows for improved imaging of concave services and difficult lesions at the risk of less accurate imaging, low field of view, and no reimbursement at present.

False-Positive Results
Although RCM has been shown to be helpful in reducing unnecessary biopsies, there still is the issue of false-positives on imaging. False-positives most commonly occur in nevi with severe atypia or when Langerhans cells are present that cannot always be differentiated from melanocytic cells.^3,15,16 One prospective study found 7 false-positive results from 63 sites using RCM for the diagnosis of lentigo malignas.¹⁶ False-negatives can occur in the presence of inflammatory infiltrates and scar tissue that can hide cellular morphology or in sampling errors due to skip lesions.^3,16

Time Efficiency
The time required for acquisition of RCM mosaics and stacks followed by reading and interpretation can be substantial depending on the size and complexity of the lesion, which is a major limitation for use of RCM in busy dermatology practices; therefore, RCM should be reserved for lesions selected to undergo biopsy that are clinically equivocal for malignancy prior to RCM examination.¹⁷ It would not be cost-effective or time effective to evaluate lesions that either clinically or dermoscopically have a high probability of malignancy; however, patients and physicians may opt for increased specificity at the expense of time, particularly when a lesion is located on a cosmetically sensitive area, as patients can avoid initial histologic biopsy and gain the cosmetic benefit of going straight to surgery versus obtaining an initial diagnostic biopsy.

Cost
Lastly, the high cost involved in purchasing an RCM device and the training involved to use and interpret RCM images currently limits RCM to large academic centers. Reimbursement may make more widespread use feasible. In any event, RCM imaging should be part of the curriculum for both dermatology and pathology trainees.

Future Directions

In vivo RCM is a noninvasive imaging modality that allows for real-time evaluation of the skin. Used in conjunction with dermoscopy, RCM can substantially improve diagnostic accuracy and reduce the number of unnecessary biopsies. Now that RCM has finally gained foundational CPT codes and insurance reimbursement, there may be a growing demand for clinicians to incorporate this technology into their clinical practice.

Reflectance confocal microscopy (RCM) imaging received Category I Current Procedural Terminology (CPT) codes by the Centers for Medicare & Medicaid Services in January 2016 and can now be submitted to insurance companies with reimbursement comparable to a skin biopsy or a global skin pathology service.¹ This fairly new technology is a US Food and Drug Administration–cleared noninvasive imaging modality that provides high-resolution in vivo cellular images of the skin. It has been shown to be efficacious in differentiating benign and malignant skin lesions, increasing diagnostic accuracy, and reducing the number of unnecessary skin biopsies that are performed. In addition to skin cancer diagnosis, RCM imaging also can help guide management of malignant lesions by detecting lateral margins prior to surgery as well as monitoring the lesion over time for treatment efficacy or recurrence. The potential impact of RCM imaging is tremendous, and reimbursement may lead to increased use in clinical practice to the benefit of our patients. Herein, we present a brief review of RCM imaging and reimbursement as well as the benefits and limitations of this new technology for dermatologists.

Reflectance Confocal Microscopy

In vivo RCM allows us to visualize the epidermis in real time on a cellular level down to the papillary dermis at a high resolution (×30) comparable to histologic examination. With optical sections 3- to 5-µm thick and a lateral resolution of 0.5 to 1.0 µm, RCM produces a stack of 500×500-µm² images up to a depth of approximately 200 µm.^2,3 At any chosen depth, these smaller images are stitched together with sophisticated software into a block, or mosaic, increasing the field of view to up to 8×8 mm². Imaging is performed in en face planes oriented parallel to the skin surface, similar to dermoscopy.

Current CPT Guidelines and Reimbursement

The CPT codes for RCM imaging provide reimbursement on a per-lesion basis and are similar to those used for skin biopsy and pathology (Table).¹ Codes 96931 through 96933 are used for imaging of a single lesion on a patient. The first code—96931—is used when image acquisition, interpretation, and report creation are carried out by a single clinician. The next 2 codes are used when one clinician acquires the image—96932—comparable to the technical component of a pathology code, while another reads it and creates the report—96933—similar to a dermatopathologist billing for the professional component of a pathology report. For patients presenting with multiple lesions, the next 3 codes—96934, 96935, and 96936—are used in conjunction with the applicable first code for each additional lesion with similar global, technical, and professional components. Because these codes are not in the radiology or pathology sections of CPT, a single code cannot be used with modifier -TC (technical component) and modifier -26, as they are in those sections.

The wide-probe VivaScope 1500 (Caliber I.D., Inc) currently is the only confocal device that can be reported with a CPT code and routinely reimbursed. The handheld VivaScope 3000 (Caliber I.D., Inc) can only view a small stack and does not have the ability to acquire a full mosaic image; it is not covered by these codes.

Images can be viewed as a stack captured at the same horizontal position but at sequential depths or as a mosaic, which has a larger field of view but is limited to a single plane. To appropriately assess a lesion, clinicians must obtain a mosaic that needs to be assessed at multiple layers for a diagnosis to be made because it is a cross-section view.

Diagnosis

Studies have demonstrated the usefulness of RCM imaging in the diagnosis of a wide range of skin diseases, including melanoma and nonmelanoma skin cancers, infectious diseases, and inflammatory and autoimmune conditions, as well as wound healing and skin aging. Reflectance confocal microscopy imaging is not limited to the skin; it can be used to evaluate the hair, nails, oral mucosa, and other organs.

According to several studies, RCM imaging notably increases the diagnostic accuracy and detection rate of skin cancers over clinical and dermoscopic examination alone and therefore can act as an aid in differentiating lesions that are benign versus those that are suspicious and should be biopsied.

Reflectance confocal microscopy has been shown to have a mean sensitivity of 94% (range, 92%–96%) and specificity of 83% (range, 81%–84%) for all types of skin cancer when used with dermoscopy.⁴ In particular, for melanocytic lesions that are ambiguous on dermoscopy, RCM used in addition to dermoscopy increases the mean sensitivity and specificity for melanoma diagnosis to 93% (range, 89%–96%) and 76% (range, 68%–83%), respectively.⁵ Although these reported sensitivities are comparable to dermoscopy, the specificity is superior, especially for detecting hypomelanotic and amelanotic melanomas, which often lack specific features on dermoscopy.^6-8

The combination of RCM with dermoscopy has reduced the number of unnecessary excisions of benign nevi by more than 50% when compared to dermoscopy alone.⁹ One study showed that the number needed to treat (ie, excise) a melanoma decreased from 14.6 with dermoscopy alone to 6.8 when guided by dermoscopy and RCM imaging.⁹ In a similar study, the number needed to treat dropped from 19.41 with dermoscopy alone to 6.25 with dermoscopy and RCM.¹⁰

These studies were not looking to evaluate RCM as a replacement test but rather as an add-on test to dermoscopy. Reflectance confocal microscopy imaging takes longer than dermoscopy for each lesion; therefore, RCM should only be used as an adjunctive tool to dermoscopy and not as an initial screening test. Consequentially, a dermatologist skilled in dermoscopy is essential in deciding which lesions would be appropriate for subsequent RCM imaging.

In Vivo Margin Mapping as an Adjunct to Surgery

Oftentimes, tumor margins are poorly defined and can be difficult to map clinically and dermoscopically. Studies have demonstrated the use of RCM in delineation of surgical margins prior to surgery or excisional biopsies.^11,12 Alternatively, when complete removal at biopsy would be impractical (eg, for extremely large lesions or lesions located in cosmetically sensitive areas such as the face), RCM can be used to pick the best site for an appropriate biopsy, which decreases the chance of sampling error due to skip lesions and increases histologic accuracy.

Nonsurgical Treatment Monitoring

One advantage of RCM over conventional histology is that RCM imaging leaves the tissue intact, allowing dynamic changes to be studied over time, which is useful for monitoring nonmelanoma skin cancers and lentigo maligna being treated with noninvasive therapeutic modalities.¹³ If not as a definitive treatment, RCM can act as an adjunct for surgery by monitoring reduction in lesion size prior to Mohs micrographic surgery, thereby decreasing the resulting surgical defect.¹⁴

Limitations

Imaging Depth
Although RCM is a revolutionary device in the field of dermatology, it has several limitations. With a maximal imaging depth of 350 µm, the imaging resolution decreases substantially with depth, limiting accurate interpretation to 200 µm. Reflectance confocal microscopy can only image the superficial portion of a lesion; therefore, deep tumor margins cannot be assessed. Hypertrophic or hyperkeratotic lesions, including lesions on the palms and soles, also are unable to be imaged with RCM. This limitation in depth penetration makes treatment monitoring impossible for invasive lesions that extend into the dermal layer.

Difficult-to-Reach Areas
Another limitation is the difficulty imaging areas such as the ocular canthi, nasal alae, or helices of the ear due to the wide probe size on the VivaScope 1500. The advent of the smaller handheld VivaScope 3000 device allows for improved imaging of concave services and difficult lesions at the risk of less accurate imaging, low field of view, and no reimbursement at present.

False-Positive Results
Although RCM has been shown to be helpful in reducing unnecessary biopsies, there still is the issue of false-positives on imaging. False-positives most commonly occur in nevi with severe atypia or when Langerhans cells are present that cannot always be differentiated from melanocytic cells.^3,15,16 One prospective study found 7 false-positive results from 63 sites using RCM for the diagnosis of lentigo malignas.¹⁶ False-negatives can occur in the presence of inflammatory infiltrates and scar tissue that can hide cellular morphology or in sampling errors due to skip lesions.^3,16

Time Efficiency
The time required for acquisition of RCM mosaics and stacks followed by reading and interpretation can be substantial depending on the size and complexity of the lesion, which is a major limitation for use of RCM in busy dermatology practices; therefore, RCM should be reserved for lesions selected to undergo biopsy that are clinically equivocal for malignancy prior to RCM examination.¹⁷ It would not be cost-effective or time effective to evaluate lesions that either clinically or dermoscopically have a high probability of malignancy; however, patients and physicians may opt for increased specificity at the expense of time, particularly when a lesion is located on a cosmetically sensitive area, as patients can avoid initial histologic biopsy and gain the cosmetic benefit of going straight to surgery versus obtaining an initial diagnostic biopsy.

Cost
Lastly, the high cost involved in purchasing an RCM device and the training involved to use and interpret RCM images currently limits RCM to large academic centers. Reimbursement may make more widespread use feasible. In any event, RCM imaging should be part of the curriculum for both dermatology and pathology trainees.

Future Directions

In vivo RCM is a noninvasive imaging modality that allows for real-time evaluation of the skin. Used in conjunction with dermoscopy, RCM can substantially improve diagnostic accuracy and reduce the number of unnecessary biopsies. Now that RCM has finally gained foundational CPT codes and insurance reimbursement, there may be a growing demand for clinicians to incorporate this technology into their clinical practice.

Reflectance confocal microscopy (RCM) imaging received Category I Current Procedural Terminology (CPT) codes by the Centers for Medicare & Medicaid Services in January 2016 and can now be submitted to insurance companies with reimbursement comparable to a skin biopsy or a global skin pathology service.¹ This fairly new technology is a US Food and Drug Administration–cleared noninvasive imaging modality that provides high-resolution in vivo cellular images of the skin. It has been shown to be efficacious in differentiating benign and malignant skin lesions, increasing diagnostic accuracy, and reducing the number of unnecessary skin biopsies that are performed. In addition to skin cancer diagnosis, RCM imaging also can help guide management of malignant lesions by detecting lateral margins prior to surgery as well as monitoring the lesion over time for treatment efficacy or recurrence. The potential impact of RCM imaging is tremendous, and reimbursement may lead to increased use in clinical practice to the benefit of our patients. Herein, we present a brief review of RCM imaging and reimbursement as well as the benefits and limitations of this new technology for dermatologists.

Reflectance Confocal Microscopy

In vivo RCM allows us to visualize the epidermis in real time on a cellular level down to the papillary dermis at a high resolution (×30) comparable to histologic examination. With optical sections 3- to 5-µm thick and a lateral resolution of 0.5 to 1.0 µm, RCM produces a stack of 500×500-µm² images up to a depth of approximately 200 µm.^2,3 At any chosen depth, these smaller images are stitched together with sophisticated software into a block, or mosaic, increasing the field of view to up to 8×8 mm². Imaging is performed in en face planes oriented parallel to the skin surface, similar to dermoscopy.

Current CPT Guidelines and Reimbursement

The CPT codes for RCM imaging provide reimbursement on a per-lesion basis and are similar to those used for skin biopsy and pathology (Table).¹ Codes 96931 through 96933 are used for imaging of a single lesion on a patient. The first code—96931—is used when image acquisition, interpretation, and report creation are carried out by a single clinician. The next 2 codes are used when one clinician acquires the image—96932—comparable to the technical component of a pathology code, while another reads it and creates the report—96933—similar to a dermatopathologist billing for the professional component of a pathology report. For patients presenting with multiple lesions, the next 3 codes—96934, 96935, and 96936—are used in conjunction with the applicable first code for each additional lesion with similar global, technical, and professional components. Because these codes are not in the radiology or pathology sections of CPT, a single code cannot be used with modifier -TC (technical component) and modifier -26, as they are in those sections.

The wide-probe VivaScope 1500 (Caliber I.D., Inc) currently is the only confocal device that can be reported with a CPT code and routinely reimbursed. The handheld VivaScope 3000 (Caliber I.D., Inc) can only view a small stack and does not have the ability to acquire a full mosaic image; it is not covered by these codes.

Images can be viewed as a stack captured at the same horizontal position but at sequential depths or as a mosaic, which has a larger field of view but is limited to a single plane. To appropriately assess a lesion, clinicians must obtain a mosaic that needs to be assessed at multiple layers for a diagnosis to be made because it is a cross-section view.

Diagnosis

Studies have demonstrated the usefulness of RCM imaging in the diagnosis of a wide range of skin diseases, including melanoma and nonmelanoma skin cancers, infectious diseases, and inflammatory and autoimmune conditions, as well as wound healing and skin aging. Reflectance confocal microscopy imaging is not limited to the skin; it can be used to evaluate the hair, nails, oral mucosa, and other organs.

According to several studies, RCM imaging notably increases the diagnostic accuracy and detection rate of skin cancers over clinical and dermoscopic examination alone and therefore can act as an aid in differentiating lesions that are benign versus those that are suspicious and should be biopsied.

Reflectance confocal microscopy has been shown to have a mean sensitivity of 94% (range, 92%–96%) and specificity of 83% (range, 81%–84%) for all types of skin cancer when used with dermoscopy.⁴ In particular, for melanocytic lesions that are ambiguous on dermoscopy, RCM used in addition to dermoscopy increases the mean sensitivity and specificity for melanoma diagnosis to 93% (range, 89%–96%) and 76% (range, 68%–83%), respectively.⁵ Although these reported sensitivities are comparable to dermoscopy, the specificity is superior, especially for detecting hypomelanotic and amelanotic melanomas, which often lack specific features on dermoscopy.^6-8

The combination of RCM with dermoscopy has reduced the number of unnecessary excisions of benign nevi by more than 50% when compared to dermoscopy alone.⁹ One study showed that the number needed to treat (ie, excise) a melanoma decreased from 14.6 with dermoscopy alone to 6.8 when guided by dermoscopy and RCM imaging.⁹ In a similar study, the number needed to treat dropped from 19.41 with dermoscopy alone to 6.25 with dermoscopy and RCM.¹⁰

These studies were not looking to evaluate RCM as a replacement test but rather as an add-on test to dermoscopy. Reflectance confocal microscopy imaging takes longer than dermoscopy for each lesion; therefore, RCM should only be used as an adjunctive tool to dermoscopy and not as an initial screening test. Consequentially, a dermatologist skilled in dermoscopy is essential in deciding which lesions would be appropriate for subsequent RCM imaging.

In Vivo Margin Mapping as an Adjunct to Surgery

Oftentimes, tumor margins are poorly defined and can be difficult to map clinically and dermoscopically. Studies have demonstrated the use of RCM in delineation of surgical margins prior to surgery or excisional biopsies.^11,12 Alternatively, when complete removal at biopsy would be impractical (eg, for extremely large lesions or lesions located in cosmetically sensitive areas such as the face), RCM can be used to pick the best site for an appropriate biopsy, which decreases the chance of sampling error due to skip lesions and increases histologic accuracy.

Nonsurgical Treatment Monitoring

One advantage of RCM over conventional histology is that RCM imaging leaves the tissue intact, allowing dynamic changes to be studied over time, which is useful for monitoring nonmelanoma skin cancers and lentigo maligna being treated with noninvasive therapeutic modalities.¹³ If not as a definitive treatment, RCM can act as an adjunct for surgery by monitoring reduction in lesion size prior to Mohs micrographic surgery, thereby decreasing the resulting surgical defect.¹⁴

Limitations

Imaging Depth
Although RCM is a revolutionary device in the field of dermatology, it has several limitations. With a maximal imaging depth of 350 µm, the imaging resolution decreases substantially with depth, limiting accurate interpretation to 200 µm. Reflectance confocal microscopy can only image the superficial portion of a lesion; therefore, deep tumor margins cannot be assessed. Hypertrophic or hyperkeratotic lesions, including lesions on the palms and soles, also are unable to be imaged with RCM. This limitation in depth penetration makes treatment monitoring impossible for invasive lesions that extend into the dermal layer.

Difficult-to-Reach Areas
Another limitation is the difficulty imaging areas such as the ocular canthi, nasal alae, or helices of the ear due to the wide probe size on the VivaScope 1500. The advent of the smaller handheld VivaScope 3000 device allows for improved imaging of concave services and difficult lesions at the risk of less accurate imaging, low field of view, and no reimbursement at present.

False-Positive Results
Although RCM has been shown to be helpful in reducing unnecessary biopsies, there still is the issue of false-positives on imaging. False-positives most commonly occur in nevi with severe atypia or when Langerhans cells are present that cannot always be differentiated from melanocytic cells.^3,15,16 One prospective study found 7 false-positive results from 63 sites using RCM for the diagnosis of lentigo malignas.¹⁶ False-negatives can occur in the presence of inflammatory infiltrates and scar tissue that can hide cellular morphology or in sampling errors due to skip lesions.^3,16

Time Efficiency
The time required for acquisition of RCM mosaics and stacks followed by reading and interpretation can be substantial depending on the size and complexity of the lesion, which is a major limitation for use of RCM in busy dermatology practices; therefore, RCM should be reserved for lesions selected to undergo biopsy that are clinically equivocal for malignancy prior to RCM examination.¹⁷ It would not be cost-effective or time effective to evaluate lesions that either clinically or dermoscopically have a high probability of malignancy; however, patients and physicians may opt for increased specificity at the expense of time, particularly when a lesion is located on a cosmetically sensitive area, as patients can avoid initial histologic biopsy and gain the cosmetic benefit of going straight to surgery versus obtaining an initial diagnostic biopsy.

Cost
Lastly, the high cost involved in purchasing an RCM device and the training involved to use and interpret RCM images currently limits RCM to large academic centers. Reimbursement may make more widespread use feasible. In any event, RCM imaging should be part of the curriculum for both dermatology and pathology trainees.

Future Directions

In vivo RCM is a noninvasive imaging modality that allows for real-time evaluation of the skin. Used in conjunction with dermoscopy, RCM can substantially improve diagnostic accuracy and reduce the number of unnecessary biopsies. Now that RCM has finally gained foundational CPT codes and insurance reimbursement, there may be a growing demand for clinicians to incorporate this technology into their clinical practice.