Optimum QMF bank based ECG data compression

Abstract

This work presents a new Electrocardiogram (ECG) data compression technique based on the optimum two channel quadrature mirror filter (QMF) bank. Firstly, prototype filter is designed using ParksMcClellan algorithm. To avoid amplitude distortion linear optimization technique is used by varying passband edge frequency of prototype filter. Than after, QMF bank is designed by optimized prototype filter. Data compression is done by decomposing the signal using optimum QMF bank and truncates the irrelevant coefficients using level thresholding. Further, Run-Length Encoding (RLE) is applied to improve the compression performance. The proposed method is tested using MIT-BIH database. The performance is estimated by different parameters viz., the compression ratio (CR), percentage root-mean-square difference (PRD), signal to noise ratio (SNR), and quality score (QS). Experimental results show that proposed method provides QS up to 20.10 and table of comparison shows that this work is better than several other existing methods in terms of CR and PRD. Diagnostic information of both original and reconstructed signals is compared, which shows both signals have same diagnostic information.