Abstracts

Rationale: Cognitive impairment is a common comorbidity in patients with temporal lobe epilepsy, and 30%–60% of patients experience verbal memory decline even after undergoing neurosurgery. Understanding brain plasticity associated with clinical phenotypes is important for effective presurgical planning. Task-based fMRI (t-fMRI) is commonly used in clinical settings to study language and memory organization in the brain during standardized tasks. Here, we computed alterations in the individual language networks of temporal lobe epilepsy patients in terms of activation level and spatial distribution using t-fMRI. Furthermore, we explored the correlation of these alterations with different clinical phenotypes, such as seizure onset lateralization, duration of epilepsy, and lesion status.

Methods: Seventy-seven temporal lobe epilepsy patients (Left = 48, Right = 29) completed a standardized sentence generation task. Patients with significantly reduced activation during the task were excluded. We computed group-level activation for the overall cohort, which served as a normative language atlas. Next, we calculated deviation maps by taking the difference between group-level and individual subject-level activation. A voxel was considered significantly altered if the difference exceeded a threshold (p < 0.01). We then counted the number of significantly deviated voxels as a measure of spatial heterogeneity. In addition, we used the mean absolute error as a measure of alterations in network activation. Finally, we correlated these two measures with seizure laterality, duration of epilepsy, and alterations across the canonical resting-state networks.