Artificial intelligence to detect multiple sclerosis

 Artificial intelligence isn't just for generating images of cute cats or creating deepfakes. In expert hands, this technology can be of great help in the field of health, for example in the early diagnosis of multiple sclerosis.

Generated with AI

It is generally at the time of the first symptoms, after an MRI, that patients suffering from multiple sclerosis or a neurogenerative disease know what to expect. This is a problem for monitoring the disease, because it is not known when or how brain structures were affected by the condition.

A bicameral parliamentary system to analyze MRIs

To identify the areas of the brain which are the first affected by a neurogenerative disease, the study of images of the brain is essential, but "  a fine segmentation of the brain on MRI images requires two weeks of work from a person specially trained for  ", explains Pierrick Coupé, research director at the CNRS. This is where artificial intelligence comes into play.

The specialist in the analysis and processing of biomedical images and his team analyzed the MRIs of 41,000 subjects, including more than 2,500 profiles from the database of the French Multiple Sclerosis Observatory. With so many brains to analyze, “  this work would simply be impossible without AI  ,” he tells the Journal du CNRS .

Its objective is to “  verify whether structures were affected early and specifically by the disease  ”. And to achieve this, researchers have developed an AI equipped with 250 neural networks that operate inspired by our way of solving problems: a bicameral parliamentary system!