Classification of Multiple Cancer Types Using Fuzzy Support Vector Machines and Outlier Detection Methods

The support vector machine (SVM) is a new learning method and has shown comparable or better results than the neural networks on some applications. In this paper, we applied SVM to classify multiple cancer types by gene expression profiles and exploit some strategies of the SVM method, including fuzzy logic and statistical theories. Using the proposed strategies and outlier detection methods, the FSVM (fuzzy support vector machine) can achieve a comparable or better performance than other methods, and provide a more flexible architecture to discriminate against SRBCT and non-SRBCT samples. Biomed Eng Appl Basis Comm, 2005(December); 17: 300-308. Received: June 6, 2005; Accepted: Sep. 22, 2005 Correspondence: Cheng-Hong Yang, Professor Department of Electronic Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung, Taiwan E-mail:[email protected] 301 BIOMEDICAL ENGINEERINGAPPLICATIONS, BASIS & COMMUNICATIONS 27 of cancer types primarily based on histological features has limitations due to their morphological similarity to other cancer types. Recently, diagnosis procedures typically involve a pathologist s interpretation with a combination of analyses, without a standardized systematic test. Accurate diagnosis would be essential for the efficacy of therapies. Under the premise of gene expression patterns as fingerprints at the molecular level, systematic methods to classify tumor types using gene expression data have been studied recently. Many scientists attempt to overcome the limitations of these conventional procedures, such as Dudoit et al. (2002), Furey et al. (2000), Golub et al. (1999), Khan et al. (2001), Mukherjee et al. (1999), Yeo and Poggio (2001), and references therein.