Introduction:
Paretic hand movements after stroke often correspond to activation in both the ipsilesional and contralesional hemispheres. Changes in intra- and interhemispheric connectivity have been reported but most studies have focused on individuals with limited motor impairment. The purpose of this pilot study was to examine connectivity between motor regions during paretic hand movement in a population with a range of motor impairment after stroke.
Methods:
Seventeen right-hand dominant individuals with hemispheric stroke (age 58.7±13.9; UE FM 34.3±12.3; days post-stroke 154.6±101.8) underwent functional MRI during paretic hand grasp/release. Data were extracted from 4 regions of interest (ROI) bilaterally: M1, dorsal premotor (PMd), supplementary motor (SMA), and visual association cortices. No participant had lesions in these cortical areas. These ROIs were included in a DCM analysis that focused on intrinsic connectivity during hand movement. Bayesian model selection was used to determine the optimal model from a set of bilateral (B), ipsilesional (IL), and contralesional (CL) models; Bayesian model averaging was used to extract coupling parameters across models.
Results:
Overall, individuals activated a bilateral network of motor regions during movement that included M1, PMd, SMA, basal ganglia, and cerebellum. There was variability between subjects as to the model that best fit the data: B for 3 subjects, IL for 6 subjects, and CL for 8 subjects. These three groups differed significantly in median UEFM scores (
p
=0.021; B=50, IL=34, CL=22) and number of blocks moved with the paretic hand in the Box & Blocks (
p
=0.03; B=35, IL=8, CL=0). Additionally, the direction and magnitude of M1-M1 coupling differed between the groups (
p
<0.016): CL M1 had a facilitatory influence on IL M1 in the B group (median=0.112) and CL group (0.469) but an inhibitory influence in the IL group (-0.524).
Conclusions:
Connectivity analysis revealed that the CL hemisphere plays a facilitatory role in some individuals and an inhibitory role in others that is related to level of motor impairment. Such variability in the role of the CL may have an impact on response to interventions aimed at altering communication between hemispheres during movement.
Identifying Groups of Determinants in Bayesian Model Averaging Using Dirichlet Process Clustering
