Electrocardiogram signal compression based on singular value decomposition (SVD) and adaptive scanning wavelet difference reduction (ASWDR) technique

In the field of biomedical, it has become necessary to reduce data quantity due to the limitation of storage in real-time ambulatory system and Tel-e-medicine system. Data compression plays an important role in this regard. Research has been underway since very beginning for the development of an efficient and simple technique for longer term benefits. This paper, therefore, presents an algorithm based on singular value decomposition (SVD) and wavelet difference reduction (WDR) techniques for ECG signal compression that deals with the huge data of ambulatory system. In particular, wavelet reduction technique has been adopted with two different scanning approaches such as fixed scan and adaptive scan of wavelet coefficients. SVD based compression techniques have great reconstruction quality with low compression rate, and WDR and adaptive scan wavelet difference reduction (ASWDR) techniques have opposite characteristics. Both the techniques boost up the performance efficiency of each other. The proposed method utilizes the low rank matrix for initial compression on two dimensional (2D) ECG image using SVD, and then WDR/ASWDR is initiated for final compression. The proposed algorithm has been tested on MIT-BIH arrhythmia record, and it was found that it is efficient in compression of different types of ECG signal with lower signal distortion based on different fidelity assessments. The evaluation results illustrate that the proposed algorithm has achieved compression rate up to 21.4:1 with excellent quality of signal reconstruction in terms of percentage-root-mean square difference as 1.7% and feature analysis of reconstructed signal for MIT-BIH Rec. 100.