Multiple sclerosis clinical forms classification with graph convolutional networks based on brain morphological...

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Multiple Sclerosis (MS) is a chronic autoimmune disease with varied clinical presentations.
• Accurate identification of MS clinical forms is crucial for patient management.
• Current identification relies on clinical evaluation and Magnetic Resonance Imaging (MRI).

Purpose of the Study:

• To develop an automated method for classifying MS clinical forms using conventional MRI.
• To leverage morphological connectome features and graph-based convolutional neural networks (CNNs).

Main Methods:

• Utilized T1-weighted MRI images from 91 MS patients in a longitudinal study.
• Applied graph-based convolutional neural networks to analyze morphological connectome features.
• Compared the proposed approach against 3D CNNs.

Main Results:

• The proposed graph-based CNN approach demonstrated high performance (F1-score) in classifying MS forms.
• Achieved superior results compared to state-of-the-art 3D CNN methods.
• The method effectively utilizes standard T1-weighted MRI data.

Conclusions:

• The developed approach offers an effective and automated solution for MS form classification.
• This method shows promise for clinical applications, including disability correlation.
• It highlights the potential of advanced AI techniques in analyzing neuroimaging data for MS.