WAILEA, MAUI – Dermatology is arguably the least technologically advanced specialty in medicine, but forces are at work to remedy that situation, according to Dr. Allan C. Halpern.
"We are this astoundingly visual specialty. Everything we do is captured in images. And there has been this astounding imaging revolution occurring all around us," said Dr. Halpern, chief of the dermatology service at Memorial Sloan-Kettering Cancer Center, New York.
"Yet, not only are we not driving the bus, most of us aren’t even sitting in one of the seats on the bus," Dr. Halpern said.
Dr. Allan C. Halpern
"Everywhere in medicine there has been application of technologies. We [skin cancer specialists] really are the slowest adopters outside of the laser/cosmetic space," Dr. Halpern noted. "If we don’t figure out how to leverage all these technologies in our astounding information age, we will get left behind in the dirt," he cautioned at the Hawaii Dermatology Seminar sponsored by Global Academy for Medical Education/Skin Disease Education Foundation.
Dr. Halpern is a key figure in a major collaborative effort to help bring dermatology into the 21st century by upgrading the state of melanoma detection. The International Skin Imaging Collaboration Melanoma Project is an ambitious new effort coordinated by Memorial Sloan-Kettering Cancer Center. Key collaborators include the National Institutes of Health cancer bioinformatics working group, the International Dermoscopy Society, the International Society for Digital Imaging of the Skin, and International Business Machines (IBM).
For the ISIC Melanoma Project, IBM intends to apply the same effort used in training its computer, Watson, to become the champion of the game show Jeopardy. But this time the challenge will be to improve methods of analyzing clinical and dermoscopic digital skin lesion images to create powerful new decision support and automated diagnostic systems, Dr. Halpern said.
The consortium already has commitments from academic institutions and industry to provide hundreds of thousands of images of melanomas and benign skin lesions accurately annotated for pathology and clinical diagnosis. This archive will be in the public domain, and will be available for teaching purposes, teledermatology, and for commercial development of automated diagnostic systems. An inventor who, for example, comes up with what he thinks is a superior smartphone app for melanoma detection can test its accuracy on 50,000 validated images, Dr. Halpern explained.
Dr. Halpern said he remains bullish about the future for smartphone melanoma apps, despite the much-publicized disappointing diagnostic accuracy of four such apps in a study published earlier this year (JAMA Dermatol. doi:10.1001/jamadermatol.2013.2382).
"That study made it to the front page of the Wall Street Journal. But the fact of the matter is it’s not surprising that the apps now are not all that accurate. Most of them are garage startups where somebody got their hands on images by going to their local dermatologist and getting a convenient sample of images," Dr. Halpern said.
By contrast, the ISIC Melanoma Project collaborative will make it possible for innovators to develop their apps and other diagnostic tools using vast sums of imaging data. Five years from now, phone apps for melanoma detection are going to be enormously improved, Dr. Halpern predicted.
"Why are these apps so important? Well, who finds most melanomas? It’s not us," he said. "If we’re not screening the public for melanoma, then how can we be negative about anything that makes the public better at finding their own melanomas?" he added.
"You hear doctors saying, ‘This is dangerous, people are going to use these things and they’re not going to go see their doctor.’ Well, right now there’s a good chance they’re not seeing their doctor anyway. And if these apps are making [individuals] dramatically more aware and more comfortable in checking their skin and coming in for a visit, that’s a good thing," Dr. Halpern asserted.
The data show that physicians detect about 15% of melanomas, while the other 85% are detected by the patients themselves, and there is strong demand for tools to help them do even better, Dr. Halpern added.
"The melanomas that many of us saw coming into our practices 30 years ago can’t be compared to the overwhelming majority of the melanomas we’re seeing in our practices today brought to our attention by patients. They’re getting much better at early detection," Dr. Halpern said.
Another major goal for the melanoma project is to develop dermatologic digital imaging standards and standardized terminology for industry in terms of camera quality, resolution, techniques, image encryption and compression, and other basic issues.
"Imagine going to get a chest x-ray in different parts of the world if every technician had their own way of doing it and there were no standards in the machines. That’s the current state of dermatologic imaging," Dr. Halpern noted.