Wrapper based Feature Selection for Virtual Colonoscopy Classification

One of the primary causes for deaths due to cancer all over the world is CRC or Colorectal cancer. In order to diagnose and treat disorders of the colon Colonoscopy is used widely. If it is performed properly it is usually safe and correct and tolerated well. The feature selection’s focus is the selection of subset variables taken from within the input to describe efficiently the data of input simultaneously bringing down the effects of variables that are not relevant as well as that of noise and still continue to give good results of prediction. An algorithm based on the socio-political change which was named ICA or the Imperialist Competitive Algorithm was introduced. K-NN or the nearest neighbor K was a classifier that maintained a minimal distance which was Euclidean between the feature of query vector and all the data in the nature of prototype training. Here the ICA was used for the selection of features in the detection of colon cancer and its treatment. Results have proved that this method is far superior to other methods in its metrics of performance.