New technology could lead to mobile apps that diagnose skin cancer
Friday, January 27th, 2017 | News
A Stanford University team has managed to create Artificial Intelligence that can correctly identify different forms of skin cancer, something that may be useful alongside more traditional methods of diagnosis in the future.
The AI was shown almost 130,000 photographs, all of which were electronically tagged with the skin disease that was present in the photo. With this information the AI learnt how to spot the signs of individual cancers. It is hoped that in the future this technology could be repacked onto mobile devices meaning the general public could perform diagnostic tests at home. The technology was repurposed from another study that aimed to teach AI how to tell the difference between cats and dogs.
After being presented with the information the AI was then tested against 21 trained skin cancer doctors to measure how accurate it is. The technology can not currently give a final diagnosis, this must be done via a biopsy, but it could be used to indicate that a biopsy might be needed. This would be especially effective as a home test on mobile phones, meaning there would be less strain on testing facilities.
There were 2,459 death from skin cancer during 2014, which is a figure that is expected to rise. Worldwide in 2012 there were an estimated 232,000 cases of malignant melanoma. It’s not one of the most aggressive forms of cancer, but it is easily preventable - so technology like this could improve survival rates even more.