Smartphones have become increasingly capable of performing a range of monitoring and detection services in recent years, whether it’s checking our eyes and ears or even scanning for cancer. We have also seen devices that look out for anemia, heart failure, and even the common cold.
So it’s fascinating to see such an approach also capable of providing accurate diagnosis of skin-cancer. A recent paper highlights how AI can accurately diagnose skin cancer, after being trained on over 130,000 images of the disease.
“We realized it was feasible, not just to do something well, but as well as a human dermatologist,” the team say. “That’s when our thinking changed. That’s when we said, ‘Look, this is not just a class project for students, this is an opportunity to do something great for humanity.’”
Put to the test
To prove its mettle, the AI was pitted against 21 certified dermatologists, and fared very well against their expert human rivals.
There are over 5 million cases of skin cancer in the US alone each year, and survival rates plummet the longer it remains undetected. Diagnosis is typically undertaken via a visual examination, both with the naked eye and a dermatoscope. If this still results in doubt, a biopsy is undertaken.
The approach undertaken by the Stanford team mirrors that of other recent breakthroughs in AI driven image analysis.
“We made a very powerful machine learning algorithm that learns from data,” the team say. “Instead of writing into computer code exactly what to look for, you let the algorithm figure it out.”
The team built upon a previous algorithm developed by Google to identify images. The algorithm was programmed to spot the difference between a malignant and benign tumor.
“There’s no huge dataset of skin cancer that we can just train our algorithms on, so we had to make our own,” the team explain. “We gathered images from the internet and worked with the medical school to create a nice taxonomy out of data that was very messy—the labels alone were in several languages, including German, Arabic, and Latin.”
The algorithm was tested across three core diagnostic tasks:
- keratinocyte carcinoma classification
- melanoma classification
- melanoma classification when viewed using dermoscopy
Across each of these, the AI was able to match the performance of the professional dermatologists. Unlike the humans, the AI can also be fine-tuned to be more or less sensitive, which can be valuable depending on what’s being assessed.
Future plans
At the moment, the application is only available via a computer, but the developers hope to eventually build a smartphone app too, which would clearly make it easier for people to take pictures of moles that may concern them and get a rapid diagnosis.
“My main eureka moment was when I realized just how ubiquitous smartphones will be,” they say. “Everyone will have a supercomputer in their pockets with a number of sensors in it, including a camera. What if we could use it to visually screen for skin cancer? Or other ailments?”
The developers believe that it will be relatively easy to transition to mobile devices, although the algorithm itself needs further testing to prove its robustness in clinical settings.
“Advances in computer-aided classification of benign versus malignant skin lesions could greatly assist dermatologists in improved diagnosis for challenging lesions and provide better management options for patients,” they conclude. “However, rigorous prospective validation of the algorithm is necessary before it can be implemented in clinical practice, by practitioners and patients alike.”
