Applying GCS networks to fuzzy discretized microarray data for tumour diagnosis

Abstract

Gene expression profiles belonging to DNA microarrays are composed of thousands of genes at the same time, representing the complex relationships between them. In this context, the ability of designing methods capable of overcoming current limitations is crucial to reduce the generalization error of state-of-the-art algorithms. This paper presents the application of a self-organised growing cell structures network in an attempt to cluster biological homogeneous patients. This technique makes use of a previous successful supervised fuzzy pattern algorithm capable of performing DNA microarray data reduction. The proposed model has been tested with microarray data belonging to bone marrow samples from 43 adult patients with cancer plus a group of six cases corresponding to healthy persons. The results of this work demonstrate that classical artificial intelligence techniques can be effectively used for tumour diagnosis working with high-dimensional microarray data. © Springer-Verlag Berlin Heidelberg 2006.