PHYCAA: Data-driven measurement and removal of physiological noise in BOLD fMRI

►We propose multivariate, data-driven method for removing physiological noise in fMRI. ►Identifies artifact via autocorrelation, power spectrum and spatial reproducibility. ►Significantly improves prediction and reproducibility of fMRI results. ►Noise shows subject/task-dependent dimensionality. ►Noise dimensionality correlated with respiratory and cardiac rate variability.