Breast Cancer Diagnosis Using Majority Voting Ensemble Classifier Approach

Abstract

One of the most common cancer affects women are the breast cancer. These are predicted using the diagnosing methodologies. The CT scans are the most common screening mechanisms is the computer aided prediction mechanism. Still, based on the reasons of genetic information, the microarray data is applied as the solution for diagnosing the cancer cells. Dealing with the microarray data it has various consequences, from those consequences one of them is its high dimensionality. The primary intention of this approach is to introduce an ensemble model to diagnose the cancer cells from the dataset. Initially, Adaptive Guided Bilateral Filter (AGBF) is applied to denoise the input images and to filter the images by sharpening them. Afterwards, to segment the images by the conversion of RGB images into grey levels the threshold mechanism is used. Minimum Redundancy Maximum Relevance (MRMR) feature inclusion and Local Search Adaptive Beta Hill Climbing (ABHC) are two hybrid feature selection approaches used in this article where the features and its importance were stored on a dataset. Then the diagnosing of the breast cancer is done by obtaining the majority voting ensemble classifier. The proposed approach is evaluated using a Breast Cancer Histopathological Image Classification (BreakHis) dataset and achieves the classification accuracy of 99.51%.