Improved lowpass differentiator for physiological signal processing

This paper describes a fractional integration-based lowpass differentiator filter for processing noisy physiological signals. The proposed scheme combines the two operations of differentiation and high frequency noise filtering into a unique noncausal finite impulse response (FIR) filter with antisymetrical impulse response. The filter coefficients, depending on the fractional order only, can be easily obtained. The designed filter is particularly advantageous in wave detection of physiological signals corrupted by high frequency noise such as electrocardiogram (ECG) and electroencephalogram (EEG) contaminated by electromyogram (EMG), and where filter specifications are not critically imposed and linear phase property is highly desired. Normalized bandwidth values were tabulated and reported for filter length selection.