A new approach for EEG feature extraction for detecting error-related potentials

摘要

Error-related negativity (ERN) is one of the electroencephalographical (EEG) traces related to the subject perception of erroneous responses. Methods for detecting the ERN from EEG signals have been widely investigated. Many known techniques use the ERN component of event-related potentials (ERP) by extracting relevant features and feeding those features to a classifier. In these approaches, feature extraction becomes the key point. In this paper, we used the t-CWT method which was based on the continuous wavelet transform (CWT) and Student's t-test for extracting ERN features. In the experiment, EEG was recorded from 5 electrode sites (Fz, FC1, Cz, FCz and FC2) during a sustianed attention experiment in 10 subjects. We used a support vector machine for classification. In addition, the algorithm provides fully automated detection and quantification methods for extracting the most discrimantive? ERP components between two cognitive states and are particular suitable for classifying single-trial ERPs. The features extracted by the algorithm can be interpreted in terms of signal characteristics that are contributing to the efficiency of classification, giving a new method for brain activity investigation.