Biomimetic Pattern Recognition in Cancer Detection

Biomimetic Pattern Recognition (BPR) is a classification process using a constructed biological structure. BPR is derived from the Principle of Homology-Continuity, which assumes members of the same class are biologically evolved and continuously connected. Recently, BPR has been successfully used in voice, facial, and iris recognition. In this article, we develop two BPR algorithms using proximity extension and two classification schemes. We investigate the performance of proposed BPR methods to detect cancer using DNA microarray data. A sample, normal or cancerous, consists of thousands of expressed genes, which are regarded as single nodes in a hyper-dimensional space. Assuming the PHC, nodes of the same class can be topologically assembled into a complex skeleton-like structure and further be covered with a tissue-layer to form a biological body. The resulting product can subsequently be used for classification. Performance for the algorithms, based on Leukemia, Bladder, Liver, and Colon cancers are studied. Our results indicate that the proposed BPR has an increase in recognition rate when compared to previous techniques. BPR has shown to be a promising approach for cancer detection using DNA microarray data.