Optimum wavelet transform-based ECG compression and dissimilarity measures on noise performance analysis

Abstract

In this study, an optimum wavelet transform-based ECG compression technique is proposed and its noise performance analysis is investigated. The major addressed issue is guaranteeing an error limit as small as possible measured by the percent root mean square difference (PRD) for the reconstructed ECG signal at every segment while keeping the compression ratio (CR) as large as possible with reasonable implementation complexity. For this purpose, an optimum wavelet transform-based compression algorithm is developed. Noise effects on the normal and the arrhythmia signal is analyzed based on the compression ratio (CR) and the reconstruction distortion. With the help of similarity measurement model it is defined that how much the reconstructed signal is similar to the original one. Two numerical metrics PRD and CR are used as the major performance evaluation parameters to analyze the results of the implemented method quantitatively. Using the developed technique, different types of orthonormal wavelets are compared.