Abstracts

Rationale: Pre-surgical language mapping is challenging for young children or those with cognitive impairment who are often unable to comply with demands of a language task during fMRI. Resting state functional connectivity (FC) quantifies co-occurring distributed brain processes and does not require task performance. We have used FC hemispheric contrast (FCHC) for different populations and found a left dominant frontal temporal language map at the group level (You et al., 2018). However, to be useful for surgical planning, FCHC needs to be validated at the individual level. Therefore, we now seek to assess this metric on an individual basis by comparing it with a semantic decision language fMRI task, Auditory Decision Description Task (ADDT). We hypothesize strong agreement between language dominance determined from FCHC compared to traditional ADDT fMRI. Methods: Participants included 46 children with epilepsy (EP: M age=14; Range 7-23 years) and 13 typically developing healthy adults (TD: M=21; Range 18-29). The average age of onset was 7.5 (SD=6; Range 0-17). Each subject completed fMRI scans (resting state & ADDT fMRI). Scans using a GE 3T scanner were acquired with 3mm3 isotropic resolution, TR=2000ms, TE=30ms, flip angle=90° FOV=192x192mm. After standard preprocessing, the data underwent denoising using the aCompCor strategy in CONN toolbox. FCHC is an experimental measure of FC within the language regions. For each vertex within gray matter, we counted vertices that were functionally connected to it from a target mask (based on Neurosynth language fMRI meta-analysis). This yielded a count for ipsilateral (Intra) and contralateral vertices (Inter). Then, the FCHC was calculated using (Intra-Inter). We determined individual level agreement for 3 conditions: ADDT activation, ADDT FCHC, and resting state FCHC. Agreement was investigated for temporal and frontal regions separately, and defined using categorical laterality of left (FCHC>0) or right (FCHC<0) and quantified using Cohen’s kappa (κ) with>.40=moderate agreement. Results: Comparing ADDT activation and ADDT FCHC, we found 62% frontal and 69% temporal region agreement for TD and 47% frontal and 57% temporal agreement for EP; κ’s ranged -.14-.24. For ADDT FCHC and rest FCHC, we found 62% frontal and 77% temporal agreement for TD and 47% frontal and 40% temporal agreement for EP; κ’s ranged -.20-.49. For rest FCHC and ADDT activation, we found 85% frontal and 62% temporal agreement for TD and 81% frontal and 50% temporal agreement for EP; κ’s ranged -.14-.45. See table 1 for within individual breakdown of laterality across group and region. Conclusions: Rest FCHC has the highest agreement with activation, compared to the other scan conditions, particularly for frontal regions in both the TD and EP groups. Disagreement was typically because FCHC maps appeared more bilateral when compared to activation. FCHC has modest agreement with activation and may not serve to replace traditional fMRI language mapping; however, the metric may be complementary. This new metric during resting state may reflect the comprehensive language network that includes homotopic areas in both hemispheres; previous work suggests that the contralateral language areas are invoked when task demands increase. Future work might prove if it is a meaningful index in other ways, such as the brain’s capacity for neural reorganization or by adding predictive value to understand postoperative outcomes. Our study is a first step in validating FCHC as a tool for predicting individual language risk for epilepsy surgery. Funding: No funding