Erin Meier¹, Swathi Kiran², Kushal Kapse³
¹Boston University, ²Boston University, ³Boston University
While the nature of language networks has been examined extensively in healthy adults, the nature of reorganized networks after brain damage has received less attention. The present project uses Dynamic Casual Modeling (DCM) to examine semantic processing networks in healthy individuals and in patients with aphasia to examine language reorganization after stroke.
Six neurologically-healthy controls and six patients with chronic aphasia completed an event-related fMRI semantic feature verification task. For DCM analysis, eight regions active across all controls were selected as voxels of interest (VOIs) while VOIs were selected for each patient due to differences in lesion size/location. Random effects Bayesian Model Search (BMS) and Bayesian Model Average (BMA) were calculated based on recommendations for clinical populations.
Analyses of full and reduced models and their corresponding connections were created for both groups. The full model analysis revealed LPCG-LFUSI and LMFG-LITG as the best fit for controls and patients, respectively. The reduced model analysis revealed LITG-LMFG as the best fit model for controls; the individual patient model analysis revealed that the best fit model was explained by the relative amount of damage in individual regions.
These results indicate that within the normal language network, frontal regions modulate inferior temporal regions during semantic processing; however, when key language nodes are removed from this network (i.e., reduced model) the dynamics of the network is altered. Notably, this altered network was qualitatively different from the actual brain damaged networks in patients, which showed modulations between intact/relatively spared regions constrained by the lesion.
Keywords: Effective Connectivity, Stroke, Aphasia