A novel self-adaptive algorithm for cancer classification based on feature reduction of SELDI-TOF data using wavelet decomposition

Surface enhanced laser desorption/ionization time of flight mass spectrometry (SELDI-TOF-MS) is one of the most powerful tool of modern proteomic technology. This high throughput mass spectrometry technology provides large number of complex data with high clinical significance. The protein sequence obtained from this technology is largely used in disease prediction, especially in cancer classification. But the high dimensionality of SELDI-TOF data presents great analytical and computational challenges for such classification using machine learning techniques. In this paper, novel technique has been proposed for dimensionality reduction of the SELDI-TOF data using wavelet decomposition. This technique is self adaptive and independent of the properties of data set. For classification purpose Support Vector Machine (SVM) has been proposed. The performance of the proposed algorithm is evaluated on ovarian and pancreatic cancer data set. The data sets were collected from National Cancer Institute, Center for Cancer Research, USA. A comparative performance analysis with another recently reported algorithm in literature reveals that our method can reduce the dimensionality of the data set more effectively with improved classification accuracy, sensitivity and specificity.