Validating atlas-based lesion disconnectomics in multiple sclerosis: A retrospective multi-centric study.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Image Analysis

Background:

• Magnetic resonance (MR)-based connectivity modeling is hindered by the need for advanced diffusion imaging, not standard in clinical settings.
• Tractography algorithms, commonly used for brain connectivity analysis, are sensitive to white matter lesions and acquisition parameters, limiting clinical application in diseases like multiple sclerosis (MS).

Purpose of the Study:

• To develop and validate an atlas-based approach for assessing structural disconnectivity and its relationship with lesions in MS, bypassing the requirement for individual diffusion imaging.
• To compare the proposed atlas-based method with traditional tractography-derived disconnectomes.
• To explore the clinical utility of the atlas-based approach in correlating radiological findings with clinical symptoms in MS.

Main Methods:

• An atlas-based computational pipeline was developed to study brain structural disconnectivity without individual diffusion imaging.
• The approach was validated across three multi-center multiple sclerosis datasets with varying MR field strengths (1.5 T and 3 T).
• Topological graph properties were analyzed to compare atlas-based disconnectomes with those derived from individual tractography.

Main Results:

• Atlas-based disconnectomes served as effective approximations of individual disconnectomes derived from diffusion imaging.
• A decrease in small-worldness was observed with increasing total lesion volume, indicating reduced brain connectivity efficiency in MS patients.
• Global efficiency and total lesion volume successfully stratified MS patients into subgroups with distinct clinical scores across all cohorts.

Conclusions:

• The proposed atlas-based method offers a viable alternative for studying brain connectivity in MS, overcoming limitations of traditional diffusion imaging and tractography.
• This approach demonstrates potential for bridging the gap between neuroimaging findings and clinical manifestations in multiple sclerosis.
• The findings highlight the utility of graph theory metrics applied to atlas-based connectomes for patient stratification and understanding disease impact in MS.