Individual-specific resting-state networks predict language dominance in drug-resistant epilepsy

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Accurate mapping of individual-specific cortical networks is crucial for understanding brain function, especially in neurological disorders like drug-resistant epilepsy.
• Current methods often rely on group-average data, which may not fully capture individual variations in functional organization.

Purpose of the Study:

• To reliably estimate individual-specific resting-state cortical networks in patients with drug-resistant epilepsy.
• To determine if language network topography derived from resting-state functional magnetic resonance imaging (fMRI) can predict task-based language dominance.

Main Methods:

• Utilized a multisession hierarchical Bayesian model (MS-HBM) trained on drug-resistant epilepsy patients (n=65) using 6-24 minutes of resting-state fMRI.
• Compared MS-HBM performance against models trained on healthy participants (n=40) and validated generalizability in an independent epilepsy cohort (n=26).
• Assessed the prediction accuracy of resting-state language network topography for task-based language dominance.

Main Results:

• MS-HBM trained on epilepsy patients yielded more accurate individual-specific cortical networks compared to group-average networks or models trained on healthy individuals.
• The model demonstrated generalizability in an independent cohort, with evoked fMRI activity aligning better with individual-specific networks.
• Individual-specific language network topography accurately predicted task-based language dominance (AUCs ranging from 0.72 to 0.83).

Conclusions:

• The MS-HBM effectively captures functional network reorganization in drug-resistant epilepsy.
• This approach enables accurate, individual-level prediction of language lateralization.
• The findings have direct implications for improving presurgical functional mapping in epilepsy patients.