Pathway Anchored Multimodal Clustering Reveals Circuit Level Signatures in Parkinsons Disease

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Parkinson's disease (PD) is increasingly recognized as a disorder affecting interconnected brain circuits.
• Current neuroimaging analyses often overlook the importance of specific pathway structures within these circuits.

Purpose of the Study:

• To develop and validate a novel pathway-anchored, multimodal clustering framework for analyzing brain imaging data in Parkinson's disease.
• To create an interpretable summary measure, the Multimodal Pathway Integrity Score (MPIS), for assessing imaging integrity within specific brain circuits.

Main Methods:

• Integration of structural MRI, diffusion MRI, and DAT-SPECT data using Scalable Robust Variational Compositional Co-clustering (SRVCC) within anatomically defined circuits.
• Derivation of MPIS by aggregating normalized volume, microstructural, and dopaminergic measures for each pathway.
• Analysis of the Parkinson's Progression Markers Initiative (PPMI) cohort with stability checks and covariate adjustments.

Main Results:

• SRVCC successfully identified stable imaging-derived patient clusters and feature modules in the PPMI cohort.
• MPIS demonstrated associations between reduced nigrostriatal/frontostriatal integrity and higher motor burden (UPDRS-III).
• Decreased sensory/visuospatial and limbic integrity correlated with lower global cognition (MoCA), while microvascular markers stratified imaging profiles but showed limited cross-sectional coupling.

Conclusions:

• The pathway-aware framework provides a principled and reproducible method for summarizing multimodal imaging data in Parkinson's disease.
• MPIS offers a valuable tool for understanding structure-function relationships in PD circuits.
• This approach may facilitate future research in circuit-informed patient stratification, prognosis, and the development of targeted outcome measures.