Comparison of the neural-network-based adaptive filtering and wavelet transform for R, T and P waves detection

An electrocardiogram is quite an important tool to find out more information about the heart. Despite the presence of serious noise, reliable detection of the QRS complex, and T and P waves is essential for an exact ECG analysis system. ECG analysis methods can be divided into three functional groups: direct methods, transformation methods, and parameter estimation methods. For real-time processing nonsyntactic, neural network based detection is ideal. Both of the selected methods accomplish filtering of the ECG signal. Neural network based adaptive matched filtering is capable of learning and becoming time-varying. These filters estimate the deterministic signal and remove uncorrelated noise with the deterministic signal. This method can produce better results than nonparametric algorithms. These advantages result in lower noise sensitivity and required sampling rate and accuracy at the same detection rate for all the ECG characteristic points