Gradient artefact modelling using a set of sinusoidal waveforms for EEG correction during continuous fMRI

Abstract: Simultaneous use of EEG and fMRI has become a powerful and attractive multimodal brain imaging technique in recent years, and has been broadly employed in neuroscience research as well as in clinical practice. The possibility to integrate the high temporal resolution of EEG with the high spatial resolution of fMRI constitutes one of the relevant advantages of using combined EEG-fMRI. Meanwhile, a number of challenges have to be overcome in order to consolidate such a technique as an independent and effective method to brain imaging. In particular, the artefacts which arise in the EEG signal, induced by varying gradient magnetic fields within the fMRI magnetic scanner. This work presents a novel methodology for waveform artefact modelling in order to subtract and clean up the EEG recordings. Implementation of our method results from the combination of two techniques: a non-linear low-pass filter approach based upon the slope adaption between consecutive samples of the signal (SSD); and the modelling proposal for the underlying gradient artefact components which is implemented using the sum of a set of sinusoid waveforms. The resulting EEG restoration obtained by our methodology shows to be similar to the artefact correction achieved by the established average artefact subtraction (AAS) method. Moreover, the proposed mathematical model for the artefact components shows to predict the variability of the artefact waveform over the time.