Denoising physiological data collected during multi-band, multi-echo EPI sequences

Abstract

Collecting physiological data during functional magnetic resonance imaging (fMRI) experiments can improve fMRI data cleaning and contribute to our understanding of psychophysiological processes; however, these recordings are frequently fraught with artifacts from the MRI pulse sequence. Here, we assess data from BIOPAC Systems, Inc., one of the most widely used manufacturers of physiological monitoring equipment, evaluate their recommendations for filtering such artifacts from electrocardiogram and electrodermal activity data collected during single-band, single-echo fMRI sequences, and extend these recommendations to address artifacts associated with multi-band, multi-echo fMRI sequences. While the magnitude and frequencies of artifacts differ with these aspects of pulse sequences, their effects can be mitigated via application of digital filters incorporating slice collection, multi-band factor, and repetition time. The implementation of these filters is provided both in interactive online notebooks and an open-source denoising tool.