Modeling MR induced artifacts contaminating electrophysiological signals recorded during MRI

The purpose of this paper is to present a novel parametric model for Magnetic Resonance (MR) induced artifacts contaminating electrophysiological signals (ECG, EEG, EMG, etc.) recorded simultaneously during MRI. The aim to construct an analytical representation of these artifacts is of great importance as it helps to understand and make appropriate hypotheses about the artifacts' generation process. The model presented in this paper assumes a periodic and stationary nature of these artifacts. Statistical KPSS tests were applied to confirm that observed artifacts are weak-sense stationary. The model based on a sum of sinusoids of different amplitudes, frequencies and phase delays {A, f, Φ} was most suited to represent these artifacts. The sinusoidal model parameters {A, f, Φ} were estimated by BFGS optimization. The lowest mean square error (MSE) is used to determine the model with the optimum parameters. Pearson's correlation coefficients were used as indices to evaluate the accuracy of the calculated model.