Spatial overlap of language networks from resting state fMRI in patients with epilepsy

Authors

• R Rolinski
• X You
• J Gonzalez-Castillo
• R Reynolds
• SK Inati
• WH Theodore

Abstract

Resting state (rsfMRI) may supplement task-related functional MRI (fMRI) for preoperative language mapping in epilepsy patients considered for surgery. During rsfMRI, participants lie quietly; changes in baseline blood oxygen-level dependent signal are measured, reducing need for patient compliance. Regions of interest (ROIs) chosen from prior data are used in correlation analysis to identify networked regions. ROI selection method may change network interpretation, affecting pre-surgical planning. Three rsfMRI runs and two fMRI language tasks were collected from 34 patients referred to NIH for pre-surgical evaluation (mean age, 33 years; SD 12; 12 female). ROIs were generated from peak fMRI activations during language tasks at single subject and group levels, and cortical atlas regions traditionally associated with language: inferior frontal (Broca’s area) and superior temporal gyrus (Wernicke’s area). Whole brain correlation analysis produced functional connectivity language network maps. Dice Similarity Coefficient (DSC) measured pairwise spatial overlap, quantifying similarity among language networks produced from all ROIs within patients and across 3 rsfMRI runs. At significant thresholds (correlation coefficient r=0.20, p0.60), moderate spatial overlap between task and atlas ROIs (mean DSC>0.40), and fair spatial overlap between task ROIs (mean DSC>0.20). Language networks show moderate similarity across ROI selection methods for rsfMRI analysis within individual patients. Greatest similarity was found with atlas ROIs, indicating atlas ROIs may be sufficient to identify language networks in rsfMRI analysis. Resting state fMRI is a potential pre-surgical planning tool for patients with drug-resistant focal epilepsy.

Scientific Focus Area: Neuroscience

This page was last updated on Friday, March 26, 2021