A Machine Learning-Based Approach to Oral Cancer Detection

About one-third of all occurrences of oral cancer worldwide occur in India, making it a serious health problem. High intake of tobacco and betel quid, particularly in smokeless versions, as well as alcohol consumption and inadequate dental hygiene are the main contributing factors. Due to restricted access to healthcare, diagnosis is frequently made later in life, making treatment more difficult. While government programs concentrate on early detection and tobacco control, education and awareness campaigns are still essential to lowering the incidence of oral cancer in the nation.A major public health concern is oral cancer, particularly in developing nations like India where poor survival rates are frequently caused by delayed detection. Early detection is important yet difficult because of healthcare access issues and the use of manual testing techniques. Deep learning advances have made automatic detection a viable option. In order to diagnose oral cancer from medical photos, this research analyses the accuracy of three well-known architectures: VGG16, CNN, ResNet101,DesNet201 and EfficientNet. Our goal is to improve early diagnosis and advance the development of AI-driven diagnostic tools by determining the most effective model. The rising incidence of oral cancer, especially in emerging nations, calls for creative approaches to prompt identification. Poor patient outcomes and treatment delays are frequently caused by the inaccuracies and inefficiencies of traditional diagnostic techniques. To diagnose oral cancer from medical photos, this research intends to investigate and compare the performance of three different deep learning architectures: VGG16, CNN, ResNet101, DesNet201 and EfficientNet. We hope to add to the expanding corpus of knowledge on AI applications in healthcare by methodically assessing these models, which will eventually enable early therapies and raise the survival rates of patients with oral cancer.