Modeling MR induced artifacts contaminating electrophysiological signals recorded during MRI

Area of Science:

• Biomedical Engineering
• Medical Imaging Physics
• Signal Processing

Background:

• Magnetic Resonance Imaging (MRI) is widely used for diagnostics.
• Simultaneous recording of physiological signals (ECG, EEG, EMG) during MRI is crucial for comprehensive patient monitoring.
• MRI poses a significant challenge due to induced artifacts that contaminate electrophysiological recordings.

Purpose of the Study:

• To develop a novel parametric model for quantifying Magnetic Resonance (MR) induced artifacts in electrophysiological signals.
• To establish an analytical representation of these artifacts to understand their generation process.
• To provide a tool for mitigating or correcting these artifacts in simultaneous recordings.

Main Methods:

• Assessed the periodic and stationary nature of MR artifacts using statistical KPSS tests.
• Developed a parametric model based on a sum of sinusoids with varying amplitudes, frequencies, and phase delays.
• Employed BFGS optimization for estimating sinusoidal model parameters.
• Utilized Mean Square Error (MSE) to select optimal model parameters and Pearson's correlation coefficients for accuracy evaluation.

Main Results:

• Confirmed the weak-sense stationary nature of observed MR artifacts.
• Demonstrated that a sum of sinusoids effectively represents the MR artifacts.
• Successfully estimated model parameters {A, f, Φ} using BFGS optimization.
• Achieved high accuracy in modeling artifacts as indicated by Pearson's correlation coefficients.

Conclusions:

• The proposed sinusoidal parametric model accurately represents MR-induced artifacts in electrophysiological signals.
• The model provides a valuable tool for understanding artifact generation and for developing artifact reduction strategies.
• This work facilitates more reliable physiological signal analysis during simultaneous MRI acquisition.