Structural Connectivity Analysis of Stroke Lesions Causing Hemiparesis : Lesion-to-Symptom Mapping Using Normative Diffusion Tensor Imaging

Abstract

Patients with brain lesions offer a unique opportunity to investigate the brain localizations of symptoms. In recent years, the interest in white-matter disconnections underlying neurological and psychiatric symptoms has grown, which has facilitated the development of new tools to investigate normative lesion-connectivity. However, as these tools have only recently been developed, their applicability for clinical lesion data is not fully known. In this project, the clinical applicability of BCBtoolkit, the first software specifically developed for structural connectivity analysis, was tested using brain lesion data from patients with post- stroke hemiparesis, as hemiparesis is expected to localize to the contralateral corticospinal tract (CST). Firstly, BCBtoolkit allows for atlas-based estimations of white-matter disconnections (BCB tractotron). Secondly, the software calculates connectome-based connectivity maps for lesions (BCB disconnectome maps) and on that basis, identifies clusters significantly associated with symptom scores (BCB AnaCOM2). The dataset in- cluded 33 cases with and 134 cases without hemiparesis. BCB tractotron analysis revealed that hemiparesis was significantly associated with the contralateral CST but also highlighted multiple other white-matter tracts. BCB AnaCOM2 analyses, testing disconnectome-based disconnections, resulted in the localization of hemiparesis to the contralateral CST with only minor anatomical unspecificity. These findings demonstrate that analyses of both atlas-based and disconnectome-based measures from BCBtoolkit identified the correct neural substrate of hemiparesis, but disconnectome-based analyses (BCB AnaCOM2) resulted in more specific anatomical associations than the analysis of atlas-based disconnections (BCB tractotron). This suggests a better applicability for clinical lesion data with a disconnectome- based analysis like BCB AnaCOM2 and alternative statistical methods may be explored in future studies.