Analysis of EEG Using Principal Component Approach

• Published in: 2007 14th IEEE International Conference on Electronics, Circuits and Systems pp. 134 - 137
• Main Authors: Padmasai, Y., Subba Rao, K., Raghavendra Rao, C., Sita Jayalakshmi, S.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2007
• Subjects: Biomedical signal processing; Chi-Square; Covariance matrix; Electrodes; Electroencephalography; Epilepsy; Multidimensional signal processing; Performance index; principal component; Principal component analysis; Scalp; Signal processing algorithms; Testing
• ISBN: 142441377X; 9781424413775
• DOI: 10.1109/ICECS.2007.4510948

The recognition of epileptic waveforms from the electroencephalogram is an important physiological signal processing task, as epilepsy is still one of the most frequent brain disorders. The main goal of this paper is to present a method to detect the epileptic waveforms directly from EEG, by performing a quick statistical signal processing. Here the principal component analysis (PCA) technique is used for reducing multidimensional data sets to lower dimensions for simplifying the analysis. In order to evaluate the statistical significance of the classification obtained in the experimental part, the Chi-square test was applied to the results. The two coefficients of significance, viz. Cramer's (V) and Sakoda's adjusted Pearson's C, (C*) are also computed. Based on the analysis of the experimental results it can be concluded that C* measure yields better results than other measures. The histograms drawn based on the PCA clearly distinguished normal subjects from abnormal ones when C* approached one.