Reproducibility of resting-state functional connectivity in healthy aging and brain injury: A mini-multiverse analysis

Resting-state functional connectivity (RSFC) methods are the most widely applied tools in the network neurosciences, but their reliability remains an active area of study. We use back-to-back 10-min resting-state scans in a healthy aging (<i>n</i> = 41) and traumatic brain injury (TBI) sample (<i>n</i> = 45) composed of older adults to assess the replicability of RSFC using a "mini" multiverse approach. The goal was to evaluate the reproducibility of commonly used graph metrics and determine if aging and moderate-severe TBI influences RSFC reliability using intraclass correlation coefficients (ICCs). There is clear evidence for reliable results in aging and TBI. Global network metrics such as within-network connectivity and segregation were most reliable whereas other whole-brain connectivity estimates (e.g., clustering coefficient, eigenvector centrality) were least reliable. Analysis of canonical networks revealed the default mode and salience networks as most reliable. There was a notable influence of motion scrubbing on ICCs, with diminished reliability proportional to the number of volumes removed. Choice of brain atlas had a modest effect on findings. Overall, RSFC reproducibility is preserved in older adults and after significant neurological compromise. We also identify a subset of graph metrics and canonical networks with promising reliability.