Rider-chicken optimization dependent recurrent neural network for cancer detection and classification using gene expression data
Chetan Nimba Aher, Ajay Kumar Jena
Abstract
One of the deadly diseases prevailing worldwide is cancer. The rigorous symptoms of cancers should be studied properly prior to the diagnosis to save patients life. Thus, an automatic prediction system for classifying cancer using gene expression data is needed. This paper develops a cancer classification and detection method by proposing the Rider Chicken Optimisation algorithm dependent Recurrent Neural Network (RCO-RNN) classifier. At first, pre-processing is done on the gene expression data to fit for the further processes of classification. In gene selection, the genes are selected based on entropy for reducing the dimension. Finally, the selected genes are classified using Recurrent Neural Network (RNN), which is trained by using the proposed Rider Chicken Optimisation (RCO) algorithm, which is the integration of Chicken Swarm Optimisation (CSO), and Rider Optimisation algorithm (ROA). The experimentation is carried out using the Leukaemia database, Small Blue Round Cell Tumour (SBRCT) dataset and Lung Cancer Dataset. The performance of the RCO-RNN is evaluated based on specificity, sensitivity, positive predictive value (PPV), negative predictive value (NPV) and accuracy. The proposed method produces the maximal accuracy, sensitivity, PPV, NPV and specificty upto 95%. Which indicates the superiority of the proposed method.