EEG and IMU Gait Signal Processing: A Comparative Assessment of the "Reza" Exponential Filter and Classical Filters

Area of Science:

• Biomedical Engineering
• Neuroscience
• Biomechanics

Background:

• Noise significantly impacts electroencephalography (EEG) and gait signals.
• Traditional Infinite Impulse Response (IIR) filters like Butterworth, Chebyshev, and elliptic present design trade-offs.
• Evaluating novel filters against established ones is essential for signal processing optimization.

Purpose of the Study:

• To compare the performance of a novel exponential
• Reza
• filter against classical IIR filters (Butterworth, Chebyshev, elliptic).
• To assess filter efficacy for both neural (EEG) and locomotor (gait) data processing.
• To determine the impact of filter selection on signal-to-noise ratio (SNR) and power spectral density (PSD) outcomes.

Main Methods:

• Analysis of an open-source mobile brain-body imaging dataset from 49 healthy adults.
• EEG data (256-channel, 512 Hz) and IMU gait data (six APDM Opals, 128 Hz) were processed.
• Band-pass filtering (EEG: 0.5-50 Hz; IMU: 0.5-5 Hz) and statistical analysis using repeated-measures ANOVAs were performed.

Main Results:

• For EEG, filter type did not affect PSD, but Chebyshev and elliptic filters yielded significantly higher SNR than Butterworth and Reza.
• For IMU gait data, the Reza filter achieved the highest SNR, outperforming elliptic and Chebyshev filters.
• The Reza filter also preserved the most motion-band power for IMU data, followed by Butterworth, Chebyshev, and elliptic.

Conclusions:

• Filter selection critically influences EEG and gait data analysis outcomes.
• Chebyshev filters are optimal for maximizing EEG SNR, while elliptic and Reza filters offer comparable fidelity.
• For IMU gait signals, the Reza filter provides superior denoising while retaining substantial signal power, comparable to Butterworth.
• Filter choice should be tailored to the specific analytical goal (SNR maximization or band power preservation).