Factors Affecting Detection Power of Blood Oxygen-Level Dependent Signal in Resting-State Functional Magnetic...

Latest developments in magnetic resonance imaging (MRI) hardware and software have significantly improved image acquisition for functional MRI (fMRI) techniques, including resting-state fMRI (rsfMRI). Specifically, with improvements in gradient and radiofrequency coils and advances in pulse sequence designs, functional images with higher spatiotemporal resolution can be achieved. However, while smaller voxel size has the benefit of resolving finer brain structures, it also decreases voxel-wise signal-to-noise ratio (SNR) and, subsequently, temporal SNR (tSNR), which is critical for the sensitivity of fMRI. Although the improved temporal resolution allows more image frames to be collected per unit time, the ability to detect brain activity by using the high spatiotemporal fMRI has not been fully characterized. Here, we aimed to evaluate the effects of spatial smoothing, scan length, sample size, seed size, and location on resting-state functional connectivity (rsFC) and tSNR by using data from the human connectome project. Results from this analysis show an important effect of smoothing on the rsFC strength (correlation values between the seed and the target) as well as on the tSNR. In contrast, while rsFC strength is not affected by sample size, the standard error decreases with the increasing number of participants, therefore improving the detection power for larger samples. Scan length and seed size seem to have a moderate effect on rsFC strength. Finally, seed location has an important impact on rsFC maps, as rsFC strength from cortical seeds seems higher than from sub-cortical seeds. In summary, our findings show that the choice of parameters can be critical for an rsfMRI study.