The performance of ECG signals compression is influenced by many things. However, there is not a single study primarily focused on the possible effects of ECG pathologies on the performance of compression algorithms. This study evaluates whether the pathologies present in ECG signals affect the efficiency and quality of compression. Single-cycle fractal-based compression algorithm and compression algorithm based on combination of wavelet transform and set partitioning in hierarchical trees are used to compress 125 15-leads ECG signals from CSE database. Rhythm and morphology of these signals are newly annotated as physiological or pathological. The compression performance results are statistically evaluated. Using both compression algorithms, physiological signals are compressed with better quality than pathological signals according to 8 and 9 out of 12 quality metrics, respectively. Moreover, it was statistically proven that pathological signals were compressed with lower efficiency than physiological signals. Signals with physiological rhythm and physiological morphology were compressed with the best quality. The worst results reported the group of signals with pathological rhythm and pathological morphology. This study is the first one which deals with effects of ECG pathologies on the performance of compression algorithms. Signal-by-signal rhythm and morphology annotations (physiological/pathological) for the CSE database are newly published.

Department of Biomedical Engineering Faculty of Electrical Engineering and Communication Brno University of Technology Technická 12 616 00 Brno Czech Republic

Department of Physiology Faculty of Medicine Masaryk University Kamenice 753 5 625 00 Brno Czech Republic

Institute of Scientific Instruments The Czech Academy of Sciences Královopolská 147 612 64 Brno Czech Republic

International Clinical Research Center St Anne's University Hospital Brno Pekařská 53 656 91 Brno Czech Republic

References provided by Crossref.org