An AI for the accurate diagnosis of melanoma

Swedish researchers at the University of Gothenburg have developed an artificial intelligence capable of supporting medical specialists in the diagnosis of melanoma. A support to the activity of those who daily find themselves having to deal with a problem that affects the skin of patients, sometimes leading to very serious consequences if not intercepted and treated in good time.

The network neural that identifies melanoma

The initiative kicked off by feeding 937 images captured through dermatoscopes to a convolutional neural network. The algorithms thus instructed were subsequently put to the test by asking them to evaluate another 200 potential cases previously examined by specialized personnel. It has emerged an effectiveness equal to that of physicians in the flesh. The same authors of the study affirm that the technology in any case needs further tests and improvements before it can be used in the clinical setting on a large scale.

As always happens when it comes to reporting similar projects, remember that the the ultimate goal is not to replace the human factor: no AI intended for use in the medical field is designed to replace experts and professionals in the sector, but to support them in operations that for obvious reasons could give rise to errors of interpretation such as those relating to analysis of diagnostic images, also reducing the time required.

Source: Expertanswer