Early stage oral cavity cancer detection: Anisotropic pre-processing and fuzzy C-means segmentation

The high rates of oral cavity cancer incidence have been found worldwide over the past decade. The death rate from oral cavity cancer is high and increasing. This study aims to improve the tumor diagnosis accuracy in the oral cavity, duly considering image processing time. It has focused on oral Computed Tomography (CT) image pre-processing and segmentation steps to enhance image quality and clarity to improve classification result. The proposed system focused on image pre-processing and segmentation steps, using anisotropic diffusion and Fuzzy C-Means to enhance the quality of the image, then improve the accuracy of tumor detection and classification. The findings attained from the current solution are based on a proposed approach using Support Vector Machine (SVM) as the traditional machine learning method to classify the oral tumor. With the combination of the anisotropic filter and fuzzy c-means algorithm, the proposed approach achieved 90.11% accuracy, 87.5% specificity and 92.157% sensitivity rate whereas the accuracy rate of the selected current best solution is only 87.18%, This study contributes to current research mainly through the implementation of an algorithm that is able to identify small sized early tumors in image edge areas.