Reliability and validity of multi-band multi-echo fMRI

ABSTRACT Precision mapping of individual differences in human brain spontaneous activity is critical for characterizing brain function and guiding clinical translation. Multi-echo resting-state functional MRI (ME-rfMRI) has the potential to provide more reliable mapping of inter-individual differences in intrinsic brain function than conventional single-echo rfMRI (SE-rfMRI). However, the psychometric benefits of ME-rfMRI for mapping individual differences in spontaneous brain activity (SBA) have not been systematically investigated, which is fundamental for understanding the human brain in health and disease. To quantify the psychometric performance of ME-rfMRI in precisely mapping inter-individual variations in SBA, we employed a psychometric design. We scanned 27 healthy adults with both ME-rfMRI and SE-rfMRI to assess short-term (minutes) test-retest reliability, as well as validity across eyes-open (EO) and eyes-closed (EC) resting states. Our results demonstrate that ME-rfMRI improves reliability relative to SE-rfMRI by 10.8% for amplitude of low-frequency fluctuation (ALFF), 4.4% for regional homogeneity (ReHo), and 8.0% for voxel-mirrored homotopic connectivity (VMHC) in cortical regions, and by 12.5% for ALFF, 8.3% for ReHo, and 4.0% for VMHC in subcortical regions. This enhancement is achieved by increasing inter-individual variability while reducing intra-individual variability. Notably, ME-rfMRI accentuates inter-individual variability in the unimodal cortex and clearly delineates two parallel systems with distinct variability patterns within the motor cortex. Concurrently, ME-rfMRI increases effect sizes (4%-8%) for detecting individual differences between EO and EC resting states. Interestingly, ME-rfMRI reveals differential effects of eye closure on two intertwined systems in the primary motor and sensory cortices, extending previous findings. Computational simulations indicate that, compared to SE-rfMRI, ME-rfMRI facilitates experimental designs with reduced costs (i.e., 4%-11% smaller sample sizes) for precise mapping of individual differences. These findings establish, for the first time, the psychometric performance of ME-rfMRI measurements, supporting its potential utility in personalized neuroscience applications, including neurodevelopment and brain disorders.