Wavelet-neural processing for computer aided diagnosis

This paper propose to apply the wavelet transform theory in the analysis of electrocardiogram signals (ECG) for the detection of particular spikes present in the Chagas´ ECG. These signals are non-stationary, having spectral features that change in time due to unpredictable events, a fact that makes wavelet transform suitable for signal analysis and segmentation. After the wavelet preprocessing, neural networks are a useful tool for automatic classification of ECG and to provide experts with an inherent estimation of the probability of symptoms of Chagas´ infection.