Gene Expression Based Acute Leukemia Cancer Classification: a Neuro-Fuzzy Approach

In this paper, we proposed the Modified Fuzzy Hypersphere Neural Network (MFHSNN) for the discrimination of acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) in leukemia dataset. Dimensionality reduction methods, such as Spearman Correlation Coefficient and Wilcoxon Rank Sum Test are used for gene selection. The performance of the MFHSNN system is encouraging when benchmarked against those of Support vector machine (SVM) and the K-nearest neighbor (K-NN) classifiers. A classification accuracy of 100% has been achieved using the MFHSNN classifier using only two genes. Furthermore, MFHSNN is found to be much faster with respect to training and testing time.