A multi-compartment model for pathological connectomes

We present a novel microstructure-informed tractography method for estimating structural connectivity in the presence of focal pathologies, such as multiple sclerosis (MS). The model introduces a lesion compartment to accurately model the intra-axonal signal decay and refine streamline weights passing through lesions. The method was first evaluated using realistic simulations of axonal damage (44 subjects from the Human Connectome Project with simulated white matter lesions) and then tested on a dataset consisting of 84 healthy controls and 107 MS patients divided by disease phenotype. The results demonstrate that the proposed method effectively captures the pathology's impact on structural connectivity, revealing significant differences in network metrics between healthy subjects and MS patients across both datasets.