Research On ECG Signal Data Compression Method Based On Wavelet Network

In clinical diagnoses systems, a large number of ECG data are acquired during sampling. So storage and transmission of ECG data have become the problem under settlement. Therefore efficient data compression algorithm is necessary to reduce the quantity of data. Especially, in remote real-time ECG monitor, ECG's real-time compression and transmission are very important to enhance the diagnoses efficiency. It can offer reliable clinical rescue for patients.In this paper, on the basis of research on domestic and oversea ECG signals' data compression algorithms and their statuses, we make use of the wavelet transform and wavelet neural network to compress the ECG signals and acquire better effects. The main work and conclusions of this paper are shown as follows:1) We introduce the feature of ECG signal and research on various ECG data compression algorithms. Especially, we simulate and analyze some ECG data compression algorithms. On the basis of introducing neural network and wavelet transform, we make use of neural network to build model for ECG signal and wavelet transform to compress ECG data. In this paper, we use BP and RBF neural network to build system model separately, and compare the two neural networks to find the difference between them.2) The ECG compression algorithm on the basis of wavelet transform is brought forward, which uses energy concentration characteristic of wavelet coefficients. We transform ECG signal by wavelet first, and then the coefficients are processed after transform to reduce the data quantities.3) We put forward the ECG compression algorithm on the basis of wavelet and neural network. We study the wavelet neural network theory and how to build models. The wavelet networks can be used in ECG signal compression by adjusting wavelet basis and weight values. At the same time, this algorithm also can reconstruct the ECG signal very well. And the result indicates that it can reflect the location and intensity of all waves correctly. Consequently, we can reach the optimum matching for compressed and original signal. In this paper, we make use of the frequency-domain matching and OLS algorithm to fix the wavelet basis and its number. We train the wavelet network by using the adaptive gradient descending algorithms, and get good effects. We can obtain different weight vectors by using wavelet neural network in compressing different ECG signals.According to the abundant simulation, the algorithm raised in this paper can reach...