9P-E-7 Feature Selection and Classification of SELDI-TOF Mass Spectra of Hepatoma Using Gene-weighted Genetic Algorithm(Room E International session)

抄録

A classifier to classify the normal samples and the samples with hepatoma based on the sample's SELDI-TOF mass spectra is designed in this paper. A modified genetic algorithm (GA) called gene-weighted GA (GWGA) is proposed to design the classifier based on the SELDI-TOF mass spectra. To reduce the computation efforts, an approach dividing the measurement intensities within different range of m/z values into several data sectors and finding the peak intensity within each data sector is proposed. The peak intensity at each data sector is taken as features for classification. The proposed GWGA aims to select the features and minimize the number of selected features while maximize the classification accuracy. Within the GWGA, the support vector machine(SVM) is utilized as the classification approach based on the features evolved in the chromosome of GWGA.