Development of Image Classification System to Detect Oral Cancer using Machine Learning Techniques

Oral cancer is among the most prevalent serious health concerns globally and is often undiagnosed in its earlier stages; this increases the morbidity as well as mortality. Optimal treatment can only be achieved with proper diagnosis; that is, proper characterization of oral cancer with the right time. This research study reflects the ability of machine learning models on medical imaging data to detect and classify oral cancer. This paper assessed the performance of several machine learning models like KNN, Logistic Regression, Neural Network, SVM and Random Forest in categorizing oral cancer from histological images and other related diagnostic features. Our proposed neural network model performed well with an accuracy of 92.4%, precision of 92.5%, F1 score of 92.5%, Specificity of 85% and recall of 92.4%. It is indicating the suggestion of a design of advanced machine learning models, it may assist in the enhancement of the early detection system of oral cancer. In addition, these results bring into the prospects of deep learning techniques for the provision of early stage diagnosis of oral cancer as dependable support to health professionals in decision-making and serve as an improvement of patient prognosis.