Low-rank singular approximation based ECG signal compression in e-health applications

Abstract

In this paper, a compression technique for ECG signal using low-rank matrix approximation based on inter and intra beat correlation, is presented. Here, singular value decomposition (SVD) has been exploited to explore the low rank representation using truncation process that stores most significant data with few singular values. In this method, two dimensional (2-D) array of ECG signal is constructed using interpolation, zero padding and average period length. The presented compression is evaluated with MIT-BIH arrhythmia ECG signal using different fidelity parameters such as compression ratio (CR), percentage root-mean square difference (PRD), signal-to-noise ratio (SNR), and correlation (CC). The obtained results presented at different rank truncation are 4:1 to 34:1 compression ratio for signal 117. Overall results show that the efficiency of presented compression technique is good for data storage or transmission in telemedicine applications.