Discrimination of epileptic events using EEG rhythm decomposition

摘要: The use of time series decomposition into sub-bands frequency to accomplish the oscillation modes in nonstationary signals is proposed. Specifically, EEG are decomposed subbands, and most relevant them employed for detection epilepsy seizures. Since computation carried out based on Time-Variant Autoregressive model parameters, both approaches searching an optimal order studied: estimation over entire database, each database recording. feature set appraises parametric power spectral density band models. Developed dimension reduction approach high dimensional space that principal component analysis searches bands holding higher values relevance, terms performed accuracy detection. Attained outcomes k-nn classifier 29 patients reach a as 95% As result, proposed methodology provides performance when used signal. advantage interpretations may lead data, since mode can be associated with one eeg rhythms.