Electrocardiogram signal compression using singular coefficient truncation and wavelet coefficient coding

In this study, an inter- and intra-beat correlation-based compression technique for electrocardiogram (ECG) signal is proposed using singular coefficient truncation, based on singular value decomposition (SVD) and adaptive scanning wavelet difference reduction technique. In this method, correlated beat segments are arranged in N x M array, and processed with proposed compression technique. Initially, array is decomposed into triplets using SVD in which most significant energy is concentrated in first few singular values. The insignificant energy coefficients are truncated and array is retained with few significant singular coefficients. Retained array is further processed with ASDWR coding technique to achieve higher compression. The proposed technique is tested on several MIT-BIH arrhythmia ECG signal records, and performance is evaluated in form of compression ratio (CR), percentage root-mean square difference (PRD), signal-to-noise ratio and cross-correlation. The evaluation results illustrate that the proposed algorithm has achieved the CR of 24.85:1 with an excellent quality of signal reconstruction in terms of PRD as 1.89% for ECG signal Rec. 100.