2D Set Partitioning In Hierarchical Tree (SPIHT) On Embedded Devices For Multilead ECG Signal Compression

Heart disease is the number one cause of death in the world according WHO data. One of the mayor cause is the lack of cardiologists. One of the proposed solution is tele-electrocardiograph (Tele-ECG) system. However, the size of ECG data recorded from patient is not small. Therefore, it must be compressed during transmission process. This paper presents development of 2D SPIHT on embedded devices for multilead ECG signal compression. There are two main processes, preprocessing and compression process. Preprocessing includes base wander removal, beat detection, and beat normalization. Compression is started with 2D ECG array construction. Afterward, we apply wavelet decomposition to transform ECG signals from the time domain to frequency domain. Essentially, the outcome of wavelet decomposition has a similar derivative among subbands. This similarity is coded using the Set Partitioning In Hierarchical Tree (SPIHT) algorithm to reduce size of the data. The compression system is successfully implemented in an embedded device: Beagleboard-C5. In this research, we use 8 leads (Lead-1, Lead-2, and Lead-V1 to V6) ECG signals taken from Physionet database. Experiment results show that the percentage root-mean-square difference (PRD) of this system is of 1.4763 for the compression ratio of 8.