ECG quality evaluation based on wavelet multi-scale entropy

Electrocardiographic (ECG) analysis, which can be used to investigate the Cardiovascular physiological and pathological phenomena, has always been promoting new drug development and improving clinical diagnosis. With the modernization of medical treatment, evaluating the quality of ECG recordings and identifying whether they are diagnostically useful by information processing technologies become the focus of new researches. This paper presents a novel approach to appraise initial ECG recordings utilizing discrete wavelet transform (DWT) and multi-scale shannon entropy. A band-pass filter with 0.67Hz~40Hz bandwidth is proposed to preprocess the original ECG recordings and then the filtered ECG recordings are decomposed into different scales by DWT, and then after an iterative procedure based on comparing the entropy of each scale coefficient with trial threshold, the quality of ECG signals is assessed and divided into 'good quality and acceptable' (TypeI), and 'poor quality and unacceptable' (TypeII). The experimental results verify that the proposed method can effectively identify the TypeII of ECG recordings.