A new multilead ECG data compression method using Higher-Order Singular Value Decomposition

Abstract

In this paper, a new compression method based on Higher-Order Singular Value Decomposition (HOSVD) for multilead electrocardiogram (MECG) data is proposed. This method exploits intra-beat and inter-beat correlation present due to quasi-periodic nature of ECG, and inter-lead correlation present among different leads of MECG data as well. The MECG data is used to form a 3-array tensor after R-peak detection and period normalization. Then, DWT is applied to the 3-array tensor for better compression. The singular values corresponding to high frequency subbands are selected on basis of singular value ratio (SVR). The MECG data is taken from the PTB Diagnostic ECG database for evaluation of the proposed method. Simulations show that there are significant improvements in compression ratio (CR) compared to existing ECG compression methods available in the literature. Compression performance and quality of the compressed MECG data of our proposed method is assessed using CR and Percentage Root Mean Square Difference (PRD). For comparison purpose, HOSVD is directly applied to the 3-array tensor data without performing DWT.