Adaptive Bi-threshold Algorithm for ECG Compression Based on Signal Slope

Abstract

Electrocardiogram (ECG) is used to record electrical activity associated with the functioning of the heart. These signals are highly data intensive and have higher resolution; thus, ECG signals require large space for storage in database and more transmission bandwidth. The ECG signals contain information signal with some redundancies; by removing these redundancies, better ECG signal compression can be achieved. The ECG compression algorithm should have high compression ratio (CR), low percent root-mean-square difference (PRD), low reconstruction error, and less computational complexity. DCT/FFT methods use frequency transformation and parameter extraction techniques. In dynamic compression scheme, IF sampler and lossless encoder are used. Both methods require preprocessing of the ECG signal. In the proposed method, the preprocessing of ECG signal is not required; signal compression is based on two threshold values, and the noise is eliminated. Using these techniques, better CR, PRD, and less storage space are achieved.