Value and Challenges of Using the Savitzky-Golay Method for ECG Noise Reduction

• Published in: 2022 International Conference on Electrical, Computer and Energy Technologies (ICECET) pp. 1 - 5
• Main Authors: Nakajima, Isao, Muraki, Yoshiya, Ichimura, Hiroko, Morita, Seiji, Nakagawa, Yoshihide
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20-07-2022
• Subjects: Distortion; Electric potential; Electrocardiography; FFT and IFFT; Finite impulse response filters; FIR; FPGA; Moving average; Noise reduction; Physiology; Smoothing methods
• DOI: 10.1109/ICECET55527.2022.9873083

With the Savitzsky Golay method, the denoising is performed under the condition that the data interval is constant. An approximate point is produced by adding the differential coefficient of a given point to the data in a range close to that point with an appropriate weight. In this paper, the frequency response of SG and the delay of the time axis are mathematically examined, and the electromyogram mixed with the collected EMG is compared with the moving average and FIR. In SG method, the amplitude is maintained, and the temporal distortion is maintained comparison from others. The effect of ECG noise reduction and the distortion of R wave and P wave are shown with the coefficient of the polynomial as a parameter. Since the core is a weighted moving average, FPGA processing can be performed at a high speed. For actual use, an accurate RR interval is required for an electrocardiogram in which the periodic myoelectric potentials are mixed. In FPGA and DSP processing, the degree of the polynomial and the number of window frames, the two parameters of SG, must be decided at the beginning of the processing. Thus, it is necessary to carefully consider the basis for the decision. Reassessment of SG has technical and educational value in processing physiological information because there is almost no peak error on the time axis.