Volume : II, Issue : XI, November - 2013

A Fuzzy–Neural Approach for Leukemia Cancer Classification

Dr. B. B. M. Krishna Kanth

Abstract :

The classification of cancers subtypes is essential for future clinical accomplishments of microarray based cancer diagnosis. In this paper, we use the Fuzzy Hyper sphere neural network (FHSNN) classifier for the discrimination of acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) subtypes present in the leukemia dataset. Prior to classification as the number of genes are larger in number compared to the samples available in the microarray datasets hence, to find the best features(genes) for classification dimensionality reduction methods such as Signal–to–Noise Ratio, Class–Separability, Wilcoxon rank sum statistic and Fisher Ratio are used. The experimental results show that our FHSNN is able to achieve 100% accuracy with much fewer genes than the previously published methods did. In particular, amongst various systematic experiments carried out, the best classification model is achieved using a subset of features chosen by Wilcoxon rank sum statistic gene selection method. Furthermore our FHSNN is found to be much faster with respect to training and testing time.

Keywords :

Article: Download PDF   DOI : 10.36106/ijsr  

Cite This Article:

Dr.B.B.M.Krishna Kanth / A Fuzzy-Neural Approach for Leukemia Cancer Classification / International Journal of Scientific Research, Vol.2, Issue.11 November 2013

Number of Downloads : 1137

References :