An ECG Extraction and Reconstruction System with Dynamic EMG Filtering Implemented on an ARM Chip

Abstract

This paper proposes two ECG signal processing algorithms for eliminating baseline drift and electromyogram (EMG) interference. Baseline drift in ECG signals is eliminated using Moving Average Filter (MAF) and the zero-crossing segmentation method is proposed so that the reconstructed ECG signal does not have an abnormal waveform. To eliminate the EMG component, the period-squared empirical learning method of pure EMG and ECG is proposed to find the threshold for identifying EMG, and then the peripheral weighted ECG method is used to reconstruct the ECG signal. Finally, a 12-lead ECG capture and filter board design is also implemented in this paper, which can be used to verify immediate ECG filtering performance.