February 1, 2017.
A team of researchers from Stanford University in the United States has succeeded in developing software that can tell the difference between a mole and a melanoma.
A diagnosis as effective, or even better, than that of a dermatologist
Regular visits to a dermatologist allow us to know if our moles are melanomas or if they are harmless. In the near future, software will be able to detect these melanomas. It is in all that hope American researchers who have developed an artificial intelligence algorithm capable of identifying skin cancers in photos.
To allow this software to make a diagnosis as effective, or even better, than that of a dermatologist, the researchers compiled 130,000 images representing portions of skin affected by more than 2,000 different diseases. They then trained their algorithm to visually recognize a potential risk from images of suspicious moles..
Scan your moles with your smartphone
According to this work, published in the journal Nature, the researchers then asked the software to analyze more than 300 pictures of moles. Among these photos were representations of malignant melanoma. At the same time, around 20 dermatologists had to take the same exam. Results : the software’s analyzes were as accurate, if not better, than those of “ the vast majority of dermatologists “, congratulated the researchers.
This software has yet to pass clinical tests before being marketed but, eventually, the researchers hope to be able to integrate it into an application. which would allow each of us to scan our moles with our smartphone. ” If one of your moles has changed color, shape, size, thickness, has evolved, it is essential to consult a doctor quickly. », However, recalls Institut Curie.
Read also: Melanoma: researchers from Nice discover a miracle molecule
Marine Rondot
