Wavelet-based feature extraction for DNA microarray classification

Complementary DNA (cDNA) microarray-based tumor gene expression profiles have been successfully used for cancer diagnosis. The main difficulty in processing cDNA microarrays is the ultra-high dimensionality of the microarrays. In this paper, we approach the dimensionality reduction using a novel wavelet-based approach that extracts classification features through microarray-block processing, thresholding, and averaging of approximation coefficients. The proposed cancer detection system presents the extracted features to a support vector machine SVM for classification (tumor or non-tumor). To show the robustness of the proposed system, its performance is tested on two public cancer microarray databases.