Abstract
With the development of mass spectrometry technology, such as surface-enhanced laser desorption/ionisation time of flight (SELDI-ToF-MS), it has become possible to diagnose proteomic; signatures for different human cancers. Among the many research issues, feature selection and classification of proteomic patterns in serum are very typical in the discrimination of cancer patients from normal individuals. In the past several years, many machine learning methods have been proposed with encouraging results. Most of the published works, however, are based on the direct application of original mass spectra, together with dimension reduction methods like PCA or feature selection methods like T-tests. Because only the peaks of MS data correspond to potential biomarkers, it is more important to study classification methods using the detected peaks, particularly from biology point of view. This paper investigates ovarian cancer identification from the detected MS peaks by applying the Kernel Fisher Discriminant Analysis (KFDA), which has been shown to be effective for classifying high-dimensional data. Comparing with several state-of-the-art methods such as SVM and random forest KFDA gives the best performance, achieving an accuracy of 96%, sensitivity of 97% and specificity of 95% in 10-fold cross-validations.