EEG signal, as a nondestructive testing method, is play a key role in the diagnosis of brain and the functional determination of brain. Since EEG was discovered by Hans Berger in 1929, a lot of signal processing techniques have been widely applied to the analysis of clinical EEG signals. As a conventional method, Fourier transformation has been widely used for the standard quantitative analysis of the spectral decomposition of EEG signals. The validity of the Fourier technique depends on the hypothesis that the EEG signals are stationary random processes. However, in many practical applications, the simplifying assumption of EEG stationary is not satisfied due to various causes of the spontaneous brain activity under different states of the brain function, such as sleep stages, epilepto-genic transients and the changes of the physiological state of the patients. Furthermore, as we known, evoket potentials reflects event related nonstationary phenomena as both temporal variations of its mean value and temporal variations of the energies of the underlying rhythms, i.e. event related spectral perturbations.
展开▼