Could your hairdresser soon be doing your mole checks?
Is it risky or resourceful for a hairdresser to check moles?
What if you didn’t have to visit the GP every time you spotted a suspicious looking mole or skin lesion? What if all you had to do to get a diagnosis was to point and shoot your smartphone?
If researchers from the USA have their way, this could soon be a reality!
A Stanford University team has used artificial intelligence (AI) to develop a smartphone app that could be just as accurate at detecting skin cancer as a doctor.
Australia has one of the highest rates of skin cancer anywhere in the world, with an estimated two in three of us to be diagnosed with skin cancer before we turn 70. With early detection significantly reducing the risk of surgery or fatality, this new technology could be a game-changer.
Researchers developed the skin cancer program by uploading 130,000 images to a library, each representing a different skin lesion. From these images, an algorithm was made which enabled the program to compare the image library to the photo that a user has uploaded. It would then visually diagnose whether the person had a skin condition – and ultimately, whether they were at risk of skin cancer.
To test the program, researchers compared its success rate against 21 experienced dermatologists. Both the doctors and the software were required to identify malignant carcinomas and malignant melanomas and were asked whether or not they'd send a patient for a biopsy of the lesion. The AI matched the performance of the dermatologists with surprising accuracy.
Being able to have a program that can identify a potentially cancerous mole could change the way we currently diagnose skin cancer. The Stanford University research team are currently working to deliver a smartphone version of the algorithm into a user-friendly app.
"Everyone will have a supercomputer in their pockets with a number of sensors in it, including a camera," said the study's lead author, Andre Esteva in a statement published in Standford News. "What if we could use it to visually screen for skin cancer? Or other ailments?"
The Stanford researchers are hoping that this AI will one day be used to make significant contributions to visual diagnosis. In the meantime, the Cancer Council provides a detection guide listing examples of the main types of skin cancer.