Continuous Wavelet Transform: ECG Recognition Based on Phase and Modulus Representations and Hidden Markov Models
Biomedical signals are fundamental observations for analyzing the body function and for diagnosing a wide spectrum of diseases. One of the major areas where new insights can be expected is the cardiovascular domain. For diagnosis purpose, the noninvasive electrocardiogram (ECG) is of great value in clinical practice. Wavelet transforms have been applied to ECG signals for enhancing late potentials, reducing noise, QRS detection, normal and abnormal beat recognition. The chapter considers the use of local symmetry properties in automatic ECG recognition and identification by means of hidden Markov models (HMMs). The set of standard mathematical tools devoted to the use of HMMs constitutes a found theoretical basis. The chapter highlights how a suitable parameter vector, corresponding to the input observation sequence of the Markov chain, can be built and applied. It describes the local behavior of the phase of continuous wavelet transform of signals with particular points as local extrema or inflection points.