Convolutional Neural Network For Detection Of Oral Cavity Leading To Oral Cancer From Photographic Images

Abstract

Oral cancer poses a substantial global health threat, as it continues to witness escalating incidence rates and consequential mortality on a widespread scale. To enhance patient outcomes, the crucial role of early detection cannot be overlooked. This research introduces an innovative real-time approach to detect various oral cavity conditions, focusing specifically on the prediction of oral cancer using the framework of deep learning. The methodology adopted integrates the user answering questions and oral cavity images analysis, amalgamating them to improve the accuracy and reliability of our predictive model. The comprehensive questionnaires gather extensive data on dietary habits, lifestyle factors, and potential risk factors associated with oral cancer. Leveraging deep learning models such as ResNet101, ResNet50, ResNet152, and VGG19, we classify oral cavity images as either cancerous or non-cancerous. By considering the relative weightage of the questionnaire responses and image analysis predictions, we compute a final probability of oral cancer. A diverse dataset is utilized to evaluate the performance of our proposed model, assessing its accuracy, sensitivity, specificity, and overall predictive capability. The resulting system aims to provide healthcare professionals with a real-time prediction tool featuring a user-friendly interface, thereby facilitating early detection and intervention. The outcomes of this study significantly contribute to the advancement of oral cancer detection methods, offering the potential to enhance patient outcomes through timely intervention.