A multiresolution time-frequency based algorithms for ECG signals compression

With the increasing use of ECG in heart diagnosis, such as 24 hours monitoring or in ambulatory monitoring systems, the volume of ECG data that should be stored or transmitted, has greatly increased. Hence, we need to reduce the data volume to decrease storage cost or make ECG signal suitable and ready for transmission through common communication channels such as mobile channel or Internet. We present in this paper two new algorithms of ECG compression. The first is based on the Modified Embedded Zerotree of Wavelet (MEZW), the second is based on the Modified Set Partitioning In Hierarchical Tree (MSPIHT). In both algorithms, 5-level decomposition is performed to the original ECG samples. The wavelet coefficients at different subband, representing the same spatial location in the ECG samples, are loaded into a spanning tree. Significant coefficients are selected and refined progressively and encoded following the spanning tree structure. The percent Root-mean Square Difference (PRD) was adopted as measure of the distortion introduced by the compressor. Algorithms were tested on ECGs from the MIT-BIH database. Obtained results reveal the interest of this kind of wavelet-based compression algorithm.