Image processing approach to baseline drift removal from electrocardiogram signals

摘要

This paper presents a method for baseline drift removal from electrocardiogram (ECG) signals, based on digital processing of an image of ECG, composed from the one-dimensional signal. The method includes morphological operations of dilation and erosion, and additional computations. Through comparison with singular spectrum analysis, it was shown that the proposed approach can be useful both in diagnostic analysis of heart rate variability and waveform analysis of ECG complexes. Promising results have been acquired, but a weakness of the method has been identified too. Baseline drift during high amplitude changes emerging in short time frame cannot be accurately estimated. However, it is suggested that the performance of the method can be increased by further modifications of certain steps. Examples exhibited in the paper were derived from Physikalisch-Technische Bundesanstalt diagnostic ECG database.