EEG and Gait Signal Processing: Comparison between a novel filter and more traditional filtering techniques

Noise degrades both EEG and gait signals, and classical IIR filters (Butterworth, Chebyshev, elliptic) involve trade offs between passband flatness, ripple, and roll off. This study compares a novel exponential "Reza" filter with these designs for neural and locomotor data. We analyzed an open mobile brain-body dataset from 49 healthy adults (EEG: 256 channel, 512 Hz; IMUs: six APDM Opals, 128 Hz). EEG channels were grand averaged and band pass filtered at 0.5-50 Hz; IMU axes were averaged and band pass filtered at 0.5-5 Hz. Outcomes were signal to noise ratio (SNR) and power spectral density (PSD). One way ANOVAs tested the effect of filter type (Butterworth, Chebyshev I, elliptic, Reza) with Bonferroni correction (α_adj = 0.0083). For EEG, PSD did not differ among filters (p = .24). SNR differed (F(3,43)= 9.21,p = 1.45×10⁻⁵): Chebyshev yielded the highest mean SNR; elliptic and Reza were intermediate and similar to each other; both exceeded Butterworth. For IMU, SNR differed (F(3,42)= 31.69,p = 2.47×10⁻¹⁵): Reza and Butterworth were highest and not different; elliptic and Chebyshev were lower. IMU PSD also differed (F(3,42)= 171.09,p = 2.97×10⁻⁴⁴): Reza retained the most motion signal power, followed by Butterworth, with elliptic and Chebyshev retaining less. These results show that filter choice materially shapes EEG and gait outcomes. For EEG, Chebyshev maximized SNR, while elliptic and Reza maintained comparable fidelity. For IMU gait signals, Reza matched Butterworth for denoising and preserved more signal power. Findings support context specific selection of filters rather than defaulting to a single design.