Prediction of Multiple Sclerosis Patient Disability from Structural Connectivity using Convolutional Neural Networks

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Neurology

Background:

• Predicting disability progression in multiple sclerosis (MS) is crucial for patient management.
• Identifying patients who benefit from specific treatments remains a challenge.
• The relationship between brain structure and disability status in MS is not fully understood.

Purpose of the Study:

• To develop a fully automatic model for estimating the Expanded Disability Status Scale (EDSS) score in MS patients.
• To leverage brain structural connectivity for predicting disability.
• To advance the prediction of MS disease evolution.

Main Methods:

• Extracting brain structural connectivity graphs from Diffusion and T1-weighted Magnetic Resonance (MR) images.
• Combining brain grey matter parcellation and tractography to form connectivity graphs.
• Utilizing a convolutional neural network (CNN) to process connectivity data and predict EDSS scores.

Main Results:

• The proposed automated model achieved promising results in predicting EDSS scores.
• The approach demonstrates the potential of using brain structural connectivity for disability prediction.
• This work represents a significant step towards improved prediction of MS progression.

Conclusions:

• The developed CNN-based model offers a novel and automatic method for EDSS score estimation in MS.
• Brain structural connectivity is a valuable predictor of disability status in multiple sclerosis.
• This research paves the way for more accurate prognostication and personalized treatment strategies in MS management.