artificial intelligence to improve diagnostics

Tremor of the limbs. The first symptoms of Parkinson’s disease, unfortunately, do not correspond to the onset of the disease. In fact, it will be announced several years in advance. However, until now the diagnosis is exclusively clinical in nature, and treatment is symptomatic.

Now imagine a world where Parkinson’s disease could be detected long before these symptoms appeared. A world where, thanks to artificial intelligence (AI) combined with medical imaging, doctors could intervene earlier and change the course of this disease that affects almost 200,000 people in France.

Artificial intelligence – an ally of visualization?

This is the promise of researchers from the University of Technology of Troyes (UTT) Racha Subra and Ali Chkeir. Their research project, called AMPIATI (Anticipation of Parkinson’s Disease through Artificial Intelligence and Image Processing), aims to reinvent the way Parkinson’s disease is diagnosed. And this is thanks to the use of advanced medical imaging techniques and artificial intelligence to detect the disease earlier.

Today, doctors use brain imaging to confirm the presence of the disease at a late stage, in the symptomatic phase, when clinical symptoms such as hand tremors or slowness of movements are declared. The “AMPIATI” project aims to detect neurons that may have disappeared due to Parkinson’s disease at a preclinical stage, without visible symptoms.

In other words, the idea is to reveal traces of disease in the images that are invisible to the human eye. How ? By identifying what we call “biomarkers.” These indicators are used to diagnose the disease, monitor its progression and assess response to treatment. Unfortunately, to date, medical brain imaging biomarkers have not been used in clinical practice for Parkinson’s disease. This makes its diagnosis particularly dependent on symptoms. This is where the “AMPIATI” project comes into play.

So two researchers propose training artificial intelligence so that it can proactively identify traces of missing or damaged neurons, focusing on segmenting a deep region of the brain called the striatum. This area plays a critical role in various functions, including movement control. In this way, researchers can, using image processing techniques and artificial intelligence models, identify and extract biomarkers of the disease and determine the stage of its development as early as possible.

The goal of the study is to use a deep learning algorithm to detect these specific forms and discover the “biomarker-0” form. The use of these advanced technologies will enable accurate and automated assessment of disease progression status. And this is long before symptoms become obvious to patients and doctors.

The first symptoms of Parkinson’s disease appear at approximately 58 years of age. So you can imagine that a few years ago, at the age of about fifty, you took a test to identify possible signs of the disease.