Using single-subject morphological networks to elucidate the patterns of disconnection and disconnectome associated with post-stroke deficits and recovery

BACKGROUNDThe single-subject morphological network (SSMN) provides a new approach for constructing structural connectome. However, its clinical relevance in post-stroke deficits and recovery remains unexplored. METHODSThis study utilized high-resolution 3D T1-weighted images alongside behavioral and cognitive assessments across multiple domains, including language, motor, memory, and attention, collected at two weeks, three months, and one year post-stroke. The SSMN was constructed using the AAL atlas by evaluating the similarities of regional probability density derived from gray matter volume. Network disconnection and the disconnectome were evaluated by examining changes in network edges and global topological properties. The functional relevance of the SSMN was explored through its associations with post-stroke behavioral and cognitive deficits and recovery, as well as by developing machine-learning-based prediction models. RESULTSThe findings revealed that the SSMN was sensitive to post-stroke connectional and connectomal disruptions. Domain-specific disruptions in the SSMN were predictable of post-stroke deficits, with correlation pattern aligning with the neurobiological substrates of each domain. Furthermore, the predictive performance of SSMN-based models was comparable to that of other imaging modalities. Notably, normalization of the SSMN within one year post-stroke was significantly associated with functional recovery. CONCLUSIONSThese results highlight the potential of the SSMN as a novel structural imaging modality for evaluating post-stroke deficits and recovery, offering valuable insights into the neurobiological mechanisms of rehabilitation.