Optimizing RetroICor and RetroKCor corrections for multi-shot 3D FMRI acquisitions

摘要: Physiological noise, if unaccounted for, can drastically reduce the statistical significance of detected activation in FMRI. In this paper, we systematically optimize physiological noise regressions for multi-shot 3D FMRI data. First, investigate whether data are best corrected image space (RetroICor) or k-space (RetroKCor), which each segment be assigned its unique phase. addition, optimal regressor set is determined using Bayesian Information Criterion (BIC) a variety acquisitions corresponding to different contrasts and readouts. Our simulations experiments indicate that: (a) corrections more robust when performed on real/imaginary than magnitude/phase data; (b) do not outperform image-space corrections, despite ability synchronize phase acquisition time accurately; (c) model varied considerably between various techniques. These results suggest use tailored volume-wide regressors, by BIC other selection criteria, that achieves balance variance reduction potential over-fitting.