Comparison of wavelet packets with cosine-modulated pseudo-QMF bank for ECG compression

An electrocardiogram (ECG) compression technique based on subband coding has been developed in this paper to compare with compression techniques based on Wavelet Packets (WP). The filter bank designed is a cosine-modulated pseudo-QMF bank using the Kaiser Window Approach (KWA) method that does not have the perfect reconstruction (PR) property, as opposed to WP that are implemented by means of a PR filter bank. In both schemes, the coding process is the same to be able to compare the results. This is carried out using a very easy algorithm based on a thresholding technique, which provides good compression rate. Its main application is to encode long-term registers of the digitized electrocardiogram signal, in order to reduce the bit rate. To preserve the reconstructed signal accuracy, the percentage root-mean-square difference (PRD) is used as an objective measurement parameter, which is selected before compression. The tests have been done for the twelve principal cardiac leads, and the compression degree measurement is evaluated by means of the mean number of bits per sample (MBPS) and the compression ratio (CR).