Spectral feature extraction of biomedical signals

Algorithms are developed and tested to interpret EEG signals. The basis of the study is the interest doctors and biomedical engineers have in spectral feature extraction of these signals.;The work here deals with spectral modeling of brain or EEG signals. Methods based on a combination of linear prediction and homomorphic filtering are applied to simulated and real EEG signals. Another active area of research in EEG signal analysis is noise cancellation. An attempt has been made to minimise muscle noise in EEG signals using adaptive filtering. Some encouraging results are obtained with real data studies. Sequential adaptive spectral estimation of EEG signals is also studied using different adaptive algorithms.;In recent years, multidimensional spectral estimation has become an area of considerable interest. Progress has been made in the development of parametric methods for multidimensional spectral estimation in general and bispectrum estimation in particular. Here an algorithm for bispectrum estimation based on parametric models is studied and is used for detecting quadratic phase coupling in EEG signals.