Feature Fusion: An Application To Biomedical Signal Classification

Abstract

Development of a feasible support system for automating staging of neural disorder based on Electroencephalogram (EEG) is essential to speed-up diagnosis process by improving the burden of the clinician of analyzing large volume data and to accelerate large scale research. In this work Discrete wavelet transform (DWT) has been applied to extract statistically independent features and fused the features for effective classification of various EEG signal. The aim of this paper is to present a comparative study of two feature fusion approaches namely Canonical Correlation Analysis (CCA) and Discriminant Correlation Analysis (DCA). Further, our proposed method can be extended to develop a graphical user interface and promote real time implementation.