Evaluating Deep Neural Networks for Oral Cancer Prediction: A Study Using ResNet50 and DenseNet121
Ankita Sharma, Sheifali Gupta, Srinivas, Haayder M. Abbas · 2025
Globally, oral cancer which is marked by unchecked cell development in the mouth cavity poses major health hazards. Usually showing up as lumps in the mouth, it causes symptoms like changes in voice, trouble swallowing, and ongoing sores. Improving patient outcomes and survival rates calls for early detection and management. This work intends to improve oral cancer detection using modern Deep-Learning methods utilized in medical imaging. Early identification greatly improves treatment outcomes for oral cancer, a dangerous and potentially fatal disorder. In this work, we investigate the automatic categorization of oral cancer using deep CNN architectures, namely ResNet50 and DenseNet121. To give dependable support for diagnostic operations, our method entails training these models to distinguish between malignant images and those that are not. According to the experimental data, DenseNet121 performed better with great accuracy of 99% than the ResNet50 model.