Application of eigenvector estimation and SVM for EEG Signals Classification

Objective: To realize the automatic classification between melancholic and healthy persons by extracting the disease features from the melancholic's EEG signals. Methods: 1. Extracting the features from the EEG signals of melancholic and healthy persons; 2. Obtaining the characteristic parameters such as the maximum, minimum, mean and standard deviation of EEG power spectrum amplitude; 3. Training the classifier and realizing the classification based on Support Vector Machines; 4. Test and validation. Results: The present classifier, which uses power spectrum characteristic parameters extracted by eigenvector methods as classification features, has better classification accuracy comparing with the one which uses frequency feature parameters extracted by wavelet methods as classification features. It achieves the classification accuracy of 95.6%. Conclusion: This paper presented a new method for melancholia diagnose.