Very large‐scale integration architecture for wavelet‐based ECG signal adaptive coder

Abstract

Very large scale integration architecture for the blocks in the novel algorithm with parallel processing is proposed. It increases the performance of the proposed coder. Since the mean is removed from electrocardiogram (ECG) signal it is represented by less number of bits. The lifting scheme-based Haar wavelet transform with five levels decomposition is used to decompose the mean removed ECG signal. One sub-band of approximation coefficient ( A 5 ) and five sub-bands of detailed coefficients ( D 5 , D 4 , D 3 , D 2 and D 1 ) are obtained. Each sub-band is represented by the corresponding number of bits by using adaptive encoder. The converted bits are transmitted through the channel using a compact format. The coding efficiency shows that the proposed coder outperforms than other coders such as novel algorithm, Djohn, Alshamali, EZW, Benzid, Chan, Khaldi, Gurkan, Wang and Set Partitioning In Hierarchical Trees (SPIHT). The proposed coder is tested on the MIT-BIH arrhythmia database (48 records) and the MIT-BIH ECG compression test database (two records) and its performance is evaluated by using evaluation metrics such as compression ratio (CR) and per cent root mean square difference (PRD). For MIT-BIH arrhythmia database record 117, a CR of 9.132:1 is achieved by the proposed coder with PRD 1.2274%.