Lossless ECG Signal Compression Using Non-linear Predictor and ASCII Character Encoding

Abstract

Electrocardiogram is a method of recording the heartbeat of a patient electronically. Storing or transmitting enormous amount of ECG signals to another device or via online is an unendurable process without compressing the signals. The purpose of this paper is to develop an efficient Electrocardiogram (ECG) compression technique using non-linear predictor and ASCII character encoding. The digitized ECG signal values are applied to Multi-layer Perceptron (MLP) neural network algorithm for non-liner prediction and the residues of the signal are passed through ASCII character encoding for further compression. It is shown that a compression ratio of 2.3726 can be achieved through this technique without any loss of information for MIT-BIH arrhythmia database records.