HISTOPATHOLOGICAL IMAGE ANALYSIS FOR ORAL CANCER CLASSIFICATION BY SUPPORT VECTOR MACHINE

Abstract

Oral cancer is caused by the mutation of the cells in the lips or in the mouth. The incidence rate and prevalence rate of oral cancer are increasing worldwide. Recently, the Machine Learning (ML) approaches play a vital role in medical image diagnosis. They provide accurate and rapid evaluation of the analysis of histopathological images using supervised learning. In this study, three different modules are developed namely preprocessing, feature extraction and classification module. Initially, the raw histopathological image is given to the median filter for the removal of background noise in the preprocessing module. In the next module, the temporal features such as energy, entropy etc., are extracted from the color components of the filtered images. Finally, the classification is done by employing the Support Vector Machine (SVM) and K-Nearest Neighbour (KNN) to classify histopathological images as normal or abnormal. Results show that the SVM classifier is better than KNN for the classification of oral cancer. The classification accuracy on 1224 histopathological images has been improved to 98% by using SVM classifier as compared with the KNN results of 83%.