cDNA Microarray Data Based Classification of Cancers Using Neural Networks and Genetic Algorithms
Tae Seon Kim, Hyun Sung Cho, Sung Mo Jeon, Jae Woo Wee and Chong Ho Lee
School of Comp. Sci. & Elec. Eng.Catholic Univ. of Korea, KR
Keywords: cDNA microarray, gene selection, classification, cancer
In this paper, we developed intelligent gene selection and cancer classification method using genetic algorithms and neural networks. To select valid genes from gene expression data set, genetic algorithms were applied and neural networks were used as classification model to support fitness values to genetic algorithms. Compare to conventional statistical schemes, proposed genetic and artificial neural network (GA/ANN) based method can effectively find highly correlated genes from microarray data set and accurately classify the disease types. The small, round blue cell tumors (SRBCTs) which is difficult to distinguish via pathological single test was used as test disease for classification, and by using GA/ANN based optimal gene selection scheme, proposed method showed superior classification results (96% accuracy) compare to statistical method (92% accuracy). With the successful implementation of proposed method, proposed method take a role of catalyst on practical usage of microarrays on disease diagnosis.
NSTI Nanotech 2003 Conference Technical Program Abstract