An ECG compression algorithm with guaranteed reconstruction quality based on optimum truncation of singular values and ASCII character encoding. (July 2018)

• Record Type: Journal Article
• Authors: Mukhopadhyay, Sourav Kumar; Ahmad, M. Omair; Swamy, M.N.S.
• Is Part Of: Biomedical signal processing and control. Volume 44(2018)
• Page Start: 288
• Page End: 306
• Publication Date: 2018-07
• Subjects: ASCII character -- 2D ECG compression -- Optimum number of singular values -- Quality-guaranteed compression -- Singular value decomposition; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2018.05.005
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: High-performance ECG compression algorithm with guaranteed reconstruction quality. Selection of an optimum number of singular values based on a pre-defined diagnostic-feature-distortion measure. Inherent signal-encryption capability. Mean-opinion-score (MOS) of the reconstructed signal is found to be 'very good'. Abstract: An efficient electrocardiogram (ECG) compression algorithm provides two-fold benefits: first, it enlarges the storage capability and second, it enhances the transmission efficiency of the communication-link in real-time tele-monitoring applications. Maintaining the quality of the reconstructed signal at a pre-determined level is a very important criterion of an ECG compression algorithm, but the area of such quality-guaranteed ECG signal compression is still lagging behind. This paper presents a high performance quality-guaranteed two-dimensional (2D) single-lead ECG compression algorithm using singular value decomposition (SVD) and lossless-ASCII-character-encoding (LLACE)-based techniques. At the preprocessing stage, the ECG signal is de-noised, and then the noise-free signal is down-sampled if the sampling frequency of the signal is found to be higher than that of a certain threshold. Then, the ECG R-peaks are detected using a Hilbert transform (HT)-based approach to extract ECG-beats. Extracted beats are arranged to form a 2D matrix, and then the matrix is decomposed using the SVD technique. An optimum number of singular values are retained in such a way that the quality of the reconstructed ECG signal would not be degraded from a pre-defined diagnostic ECG-feature-distortion measure. The truncated right singular matrix coefficients are quantized and encoded into ASCII characters, and the truncated left singular matrix coefficients are compressed using the LLACE-based technique. The SVD and LLACE methods exploit the strong inter-beat and inter-sample correlations, respectively, of an ECG signal to attain high compression performance. Evaluation results show that the mean-opinion-score of the reconstructed ECG signals signal falls under the category 'very good' as per the gold standard subjective measure.