Comparative Study of Different Pulse Artefact Correction Techniques during Concurrent EEG-FMRI Using FMRIB

Abstract

Simultaneous acquisition of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) aims to combine the strength of both methods by investigating brain activity with the high temporal resolution of the former and the good spatial precision of the later. However, simultaneous EEG-fMRI measurements are technically challenging because both methods interfere with the other modality. The main confounding factors are the pulse artefact (PA) caused by pulsatile motion linked to the cardiac cycle and the gradient artefact (GA) produced by the temporally varying magnetic fields required for MR imaging. Both of these artefacts are orders of magnitude larger than the neuronal activity of interest, but their inherent periodicity facilitates reasonable artefact correction by post-processing techniques such as average artefact subtraction (AAS), Optimal Basis Set (OBS). While the different post processing methods for PA artefact are being used to-date, but their comparison were not studied yet. In this work, different methods - OBS, simple mean (AAS) and Gaussian-weighted mean (GWM) - implemented in open source FMRIB tool-box, were used for PA correction and to compare their performance in artefact correction while retaining the neuronal information. It has been found that, of the three different PA removal methods, OBS is better in preserving bio-signal while removing PA successfully.