How AI is Bringing a Paradigm Shift to Brain Disorders

Published Jul-16-18

Two AI approaches that can help people with Alzheimer’s and those who are having a stroke.

University of Bari, Italy

The Story:

How AI is Bringing a Paradigm Shift to Brain Disorders The use of artificial intelligence (AI) in medicine has been evolving so well in recent years that it can be incorporated into real-world clinical practice. By making use of machine learning, big data and analytics to process huge volumes of data and analyze them for patterns the technology can make predictions about patients which could result in earlier and better interventions.

One part of the body that is a major focus of these approaches is its most complicated organ - the brain. For example, AI is now capable of identifying changes in the brains of people who are most likely to develop Alzheimer's and it can do so around ten years before doctors are able to make a diagnosis. It does this by studying non-invasive MRI scans to spot any changes in how regions of the brain are connected. Medical professionals diagnose Alzheimer’s by studying symptoms i.e. when the disease may have already progressed considerably.

Machine Learning

Although there is no cure for the neurodegenerative disease (the sixth leading cause of death in the United States) earlier diagnosis can allow people to make changes to their lifestyle that could slow its progress. There are also therapeutics in development that could work better the earlier they are given.

The machine learning algorithm that can spot the brain changes was developed by scientists at the University of Bari in Italy. They trained the algorithm with MRI scans of people who had Alzheimer’s and healthy controls. The idea was to teach the algorithm to 'know' the difference between a normal and a diseased brain. The researchers then divided each brain scan into small regions and analyzed the neuronal connectivity between them.

They made no assumption about the perfect size that would allow the algorithm to make an accurate diagnosis but discovered that it was most accurate when regions being compared were about 2250 to 3200 cubic millimeters. When the algorithm was tested on a second set of scans it was able to distinguish between healthy brains and those with Alzheimer’s with 86 percent accuracy.

Artificial Intelligence and Strokes

An artificial intelligence approach can also be beneficial to patients who have had a stroke or some other brain injury. During a stroke blood flow to the brain is cut off which starves the brain cells of oxygen. They become damaged and the symptoms that follow are recognized as a stroke. One of the tools used for diagnosis in hospitals is a head CT scan and interpreting the images they produce is crucial.

Timing is critical here but often radiologists have a backlog of cases which can mean that abnormalities are missed. To the rescue are deep learning algorithms created by a company called The algorithms were trained on labeled scans from a variety of sources and are now able to detect abnormalities requiring urgent attention.

Research Activity

In addition to strokes and Alzheimer’s, artificial intelligence and machine learning may also find uses with other neurological and neurodegenerative disorders such as seizures, Parkinson’s disease and Lou Gehrig’s disease as well as neurological traumas. We are still in the very early days of AI adoption but such are the many advantages and potential benefits that more research activity is inevitable.

Share on        
Next Story »

What Can we Solve for You?