CLASSIFICATION AND DETECTION OF SINGLE EVOKED BRAIN POTENTIALS USING TIME-FREQUENCY AMPLITUDE FEATURES

The classification and detection of event-related brain potentials were investigated using statistical pattern recognition techniques. Amplitudes at sampled time points and frequency quantities have previously been used as features. Improvements to these procedures were obtained by using features from the time-frequency plane to exploit the geometric relationship between time and frequency, capitalizing on the non-stationarity of the evoked potential signals and the electroencephalogram (EEG). These features were transformed from the original data sets based upon a two-step classification/feature selection procedure which uses selected frequencies from step-1 as parameters for data filtering in step-2. Features are selected from the filtered data and bounds on the expected classification accuracy were computed for various sets of data. This system was used for classification between 2 classes of evoked potentials and for the detection of a particular single evoked potential in the electroencephalogram. A detector program was developd to operate on the test data using the predetermined feature sets selected by the two-step system. The receiver operating curves were computed indicating the detector performance and the detection accuracies were evaluated for various test data sets. Actual EEG data from human subjects participating in visual stimulation Sternberg paradigm experiments, and several artificially generated data sets were used for testing the ability of the methods to distinguish between the types of signals. The results of the new method were compared with those of previous methods using 1-step techniques, and significant improvements in classification and detection accuracies were obtained