Improved Multichannel ECG Signal Compression Based on Discrete Wavelet Transform and a Special Scanning Method

Abstract

Electrocardiography is one of basic investigative methods in cardiology through which all necessary information about electrical activity
of the heart can be collected. Recorded Electrocardiogram, ECG, signals which are huge in size, not only used for diagnosis purpose but it is
also stored for future reference. Therefore, compression of huge amount of ECG signal data is required to keep these records for observing a
patient. This paper proposes a novel approach to compress a 12-lead set of ECG waveform jointly. The proposed method provides a
reasonable trade-off between distortion and compression ratio by using a lossy compression technique. The technique exploits inter-channel
correlation of signals in cardio-complex leads by using 1-D wavelet and a special scanning method that improves compression performance.
Also we find the best threshold for wavelet coefficients via optimization algorithms. It is shown that the compression ratio of this approach
for percent root mean square difference (PRD) equal to almost 5% is higher than similar methods. At the end, we use a zip algorithm to
achieve more compression.