A-4 Establishing Test-Retest Reliability of Large-Scale Neural Networks after Neurotrauma: toward Clinically Useful Biomarkers

Abstract Objective Integrating neuropsychological assessment and resting-state functional magnetic resonance imaging (rsfMRI) may provide crucial information about brain functioning and cognition. However, rsfMRI is not ready to be used clinically due to questions about reproducibility. This study investigated the test–retest reliability of resting-state networks in moderate-to-severe TBI using back-to-back rsfMRI scans. Method 51 TBI participants and 28 healthy controls received two, 10-minute rsfMRI scans separated by minutes. The data were preprocessed with fMRIPrep. XcpEngine was used to divide the brain into 264 regions based on the Power atlas, which were then grouped into 13 brain networks. Intraclass correlation coefficients (ICCs) were calculated to examine the reliability of within-network connectivity (strength of a region’s connection to other regions within the same network) for each network across each participant’s two scans. Results ICCs for healthy controls varied in reliability (ICCs = 0.42–0.80). ICCs for the TBI group varied similarly (ICCs = 0.41–0.86). For both groups, the most reliable networks were the dorsal attention, visual, and sensory networks (ICCs = 0.77–0.81). The least reliable networks were the cerebellar and ventral attention networks (ICCs = 0.42–0.43). Overlapping intervals for the ICC values indicated non-significant differences in network reliability between the healthy control and TBI groups. Conclusions Results suggest that within-network connectivity is more reliable in attentional and sensory networks, even after significant neurological compromise. However, the variability of within-network ICCs should continue to be explored. Alterations in network reliability may relate to changes in cognition and serve as a starting point to identify resting-state biomarkers, especially after moderate-to-severe TBI.