Wavelet based neural network architecture for ECG signal compression

This paper addresses the problem of compressing Electrocardiogram (ECG) signals using the concept of adaptive sampling. The concept of adaptive sampling relates to optimum estimation of wavelet parameters that best represents a given signal. These wavelet parameters are estimated by minimizing the least mean square error between the original and approximated signal. Such an optimization approach is implemented within the frame work of neural networks by using wavelet non-linear functions in its neurons. We apply this technique for the compression of ECG signals. The experimental details of ECG compression are provided. For these experiments, the standard ECG database that was created by the American Heart Association (AHA) is used.