Non-Linear Filter for Gradient Artefact Correction during Simultaneous EEG-fMRI

Parallel to the breakthroughs on the usage of simultaneous EEG-fMRI in neurocognitive studies and research, the occurrence of artefacts in the EEG signal induced within the fMRI scanner constitutes one of the challenges to be overcome in order to broaden the range of applications of such a technique. It is the case of the gradient artefact, provoked by the variation of gradient magnetic fields. Thereby, although a number of computational methodologies have yielded a satisfactory correction of the EEG signal, novel approaches have been proposed to improve the quality of the restored EEG. This work presents a novel proposal for modelling the variability of the gradient artefact template estimated during application of the established average artefact subtraction (AAS) method. Implementation of our proposed model and its combination with the AAS method allow an effective elimination of the gradient artefact from EEG recordings. Moreover, our approach performs an adaptive removal of the underlying artefact residuals, according to the signal slope parameter characteristics.