An information theory framework for capturing multi-connectivity via spatial network encoding reveals reduced population count (Hamming weight) localized to auditory, visual, and motor networks

1. Abstract The human brain operates as a complex system where functional networks evolve and interact across spatially distributed regions. In traditional neuroimaging analyses, functional connectivity (FC), based on pairwise correlations or statistical dependencies of temporal fluctuations in the BOLD signal, has been a primary method for exploring interactions between brain regions and decoding brain function. However, traditional FC methods often overlook the intricate, multi-way interplay among brain elements that emerge from the brain’s densely interconnected nature. To overcome these limitations, we introduce a novel voxel-centric framework that captures the multi-way interactions between voxels and networks identified via high-model order independent component analysis. This framework posits that individual voxels serve as critical mediators of multi-network communication, reflecting the brain’s complex functional architecture. By encoding voxel contributions from brain networks into binary representations and quantifying the population count at each voxel via Hamming weights, the proposed method prioritizes high-contribution voxels that facilitate inter-network interactions. This approach provides new insights into the brain’s functional organization, revealing previously unrecognized patterns of voxel-to-network entanglement. Specifically, in the context of schizophrenia, our method enables the identification of spatial patterns that may underpin the cognitive and perceptual disturbances characteristic of the disorder. This enhanced understanding could improve diagnostic precision and help tailor interventions that target specific dysfunctional networks, offering a pathway to more effective treatments and better patient outcomes in schizophrenia.