Researchers have just published a study to show that a computer was better than human dermatologists at detecting skin cancer. It’s better and faster diagnosis.
The international team from Germany, the United States, and France taught an AI system to detect skin cancer. They showed it over 100,000 images and made it distinguish between malignant and benign spots on the skin.
The paper appeared in the journal Annals of Oncology.
The machine is called CNN – deep learning convolutional neural network. It was tested against 58 dermatologists from 17 countries. It was shown photos of benign moles and malignant melanomas.
The team of 58 dermatologists was separated into groups. Over 50% of the dermatologists were experts over 5 years of experience. The rest of 19% had 2-5 years of experience and 29% were beginners with only two years of experience. This is what the results of the study showed:
“Most dermatologists were outperformed by the CNN.”
The CNN has detected 95% of skin cancers, while human dermatologists detected only 86.6%. The first author of the study is Holger Haenssle of the University of Heidelberg. He stated that:
“The CNN missed fewer melanomas, meaning it had a higher sensitivity than the dermatologists,” and that it “misdiagnosed fewer benign moles as malignant melanoma … this would result in less unnecessary surgery.”
In some parts of the body – fingers, toes or scalp, melanoma is difficult to image, and AI cannot easily recognize atypical lesions of the skin.
Dermatologists’ performance got better when researchers gave them more information on the skin lesions of the patients. However, the AI is a good tool to diagnose skin cancer fast and easy. It will allow doctors to surgically remove them before it starts spreading.
According to dermatology experts Victoria Mar (Monash University in Melbourne) and Peter Soyer (University of Queensland), there are not many ways of detecting skin cancer:
“Currently, there is no substitute for a thorough clinical examination.”
Rex Austinwas born and raised in Thunder Bay Ontario on the shores of Lake Superior. Apart from running his own podcast (Ice Fishing And Other “Cool” Things), he spends his time canoeing and backpacking in Northern Ontario.. As a journalist Rex has published stories for Global News (Thunder Bay) we well as Buzz Feed and Joystiq. As a contributor to Great Lakes Ledger, Rex most covers science and health stories. Contact Rexhere