Effect of scanning duration and sample size on reliability in resting state fMRI dynamic causal modeling analysis

• DOI: 10.1016/j.neuroimage.2024.120604
• Publication Date: 2024-04-10T02:27:46Z
• Authors: Liangsuo Ma; Sarah E. Braun; Joel L. Steinberg; James M. Bjork; Caitlin E. Martin; Larry D. Keen II; F. Gerard Moeller
• Keywords: Human Connectome Project; Sample (material)

Despite its widespread use, resting-state functional magnetic resonance imaging (rsfMRI) has been criticized for low test-retest reliability. To improve reliability, researchers have recommended using extended scanning durations, increased sample size, and advanced brain connectivity techniques. However, longer runs larger sizes may come with practical challenges burdens, especially in rare populations. Here we tested if an technique, dynamic causal modeling (DCM), can reliability of fMRI effective (EC) metrics to acceptable levels without extremely long run durations or large samples. Specifically, employed DCM EC analysis on rsfMRI data from the Human Connectome Project. avoid bias, assessed four distinct DCMs gradually a randomized manner across ten permutations. We pseudo true positive false rates assess efficacy shorter (3.6, 7.2, 10.8, 14.4 minutes) replicating outcomes longest duration (28.8 min) when size was fixed at largest (n=160 subjects). Similarly, smaller (n=10, 20, …, 150 subjects) min). Our results revealed that rate below 0.05 all analyses. After reached 10.8 minutes, which yielded 92%, further extensions time showed no improvements rate. Expanding led enhanced outcomes, plateau n=70 subjects targeted top one-half ECs reference sample, regardless whether viable (10.8 employed. Encouragingly, exhibited approximately 80% n=20, 90% n=40 subjects. These suggest be option attain reliable times are not feasible.