ECG signal compression using filter bank based on Hermite polynomial.

The electrocardiogram (ECG) signal compression is necessary for storage and transmission. In this paper we propose a new filter bank based on Hermite function for the effective compression of the ECG signal. The Hermite function is used to derive the low pass filter taps and compression is carried out using discrete wavelet transform (DWT). The compression scheme is implemented and performance evaluation is presented based on the standard compression indices like compression ratio (CR), percent root mean square difference (PRD) and cross correlation coefficient (CCC). The retrieval of the dominant morphological features of the ECG waveform like P-QRS-T complex upon reconstruction are also verified. The results are presented using the MIT-BIH and CSE-DS-5 databases. The results reflect that the quality of the reconstructed signal is excellent with minimum loss of the diagnostically important features.