The Adaptive Lossless Compression Of The ECG Signals

In modern medicine science, the heart disease is a very important research project, and the ECG(electrocardiogram) signal is a very important clinical medicine information for the diagnoses and study of heart disease. For many years, lots of scholars and researchers have done a lot of work on the compression of ECG signals, and the technology and scheme of the compression are becoming more and more important. There is lot of redundancy and non-relative information in the raw ECG information data, the information lead waste of transmission and storage, and also useless. If we can exploit these redundancy and non-relative information, we can achieve the compression of the raw ECG information, and make the information doing a better work in the transmission and storage.The compression of the ECG signals can be classified into one of two categories: lossy and lossless compression. The lossy ones can achieve a very high compression ratio, but the reconstructed signals are hard to be as same as the original ones, there are errors or missing of the information. The lossless compression can not get such a high compression ratio as the lossy compression does, but the important thing is that the lossless compression can ensure there are no errors or missing of the information, the reconstructed signals are completely as same as the original ones. And the lossless compression of ECG signals is much more important than the lossy compression.The research of this paper is based on the lossless compression of ECG signals:We propose a lossless compression scheme of single channel ECG signals, which is based on the linear prediction. We design a adaptive linear predictor, which is also auto-regressive.We design a variable order adaptive RICE coding method, to achieve an efficiency compression. The coding process is dynamic, the coding order is depended on the statistical characteristic of the value of data, therefore, the coding order of every single signal can be difference, the numbers of bit of this signal can be efficient.We design a new prediction model that can adaptively analysis the correlation of inter-/intra-channel, and exploit the AR algorithm to gain the optimal estimated signals in the prediction. Moreover, the reversible K-L(Karhunen-Loeve)transform is used to reduce the redundancy of the inter-channel correlation. The transformed signals are compressed by the adaptive RICE encoder we designed.