Oral Cancer Classification Using GLRLM Combined with Fuzzy Cognitive Map and Support Vector Machines from Dental Radiograph Images

In recent years, Fuzzy Cognitive Maps (FCM) are widely used in medical decision schemes. According to data, oral cancer is the fifth most common cancer in India, with a higher occurrence among men than women. In order to save lives, correct diagnosis and prompt treatment is required. When oral cancer is identified early, the 5-year survival rate exceeds 80%, but advanced stages of the illness have 5-year survival rates that are less than 20–30%. But, the majority of cases are detected in advanced stages where treatment becomes unsuccessful. To achieve a better classification of oral cancers, Gray Level Run Length Matrix (GLRLM) is combined with Fuzzy Cognitive Map and Support Vector Machines. In this work, fifty Dental Radiograph images are used for cancer classification. Initially, Linear contrast stretching algorithm is employed to remove the noise present in the image. The filtered picture is split to create a Region of Interest (ROI). The ROI picture is used to extract the GLRLM features. The classification of cancer is based on the features extracted. From the extracted features, a fuzzy cognitive map is employed. Cancer categorization is carried through Machine Learning approaches. Measures like sensitivity, specificity, precision, accuracy, and F-score have been used to describe the performance of the SVM classifier. The study’s findings demonstrate that GLRLM-based features are strong discriminating features for statistical analysis of Dental Radiograph images and can be helpful in the identification of cancer. By combining the texture features based on GLRLM and FCM, the accuracy (94%) as a whole is improved.