Abstract
Oral Squamous Cell Carcinoma (OSCC) presents significant diagnostic challenges due to subtle initial symptoms and various risk factors, including tobacco use, HPV infection, and family history. Previous research has explored various machine learning and deep learning techniques for improving cancer detection accuracy. Traditional machine learning methods, often relying on manual feature extraction, have shown limitations in handling the complexity and variability of histopathological images. Recent advancements in deep learning, particularly Convolutional Neural Networks (CNNs), have demonstrated superior capabilities in image analysis and feature extraction, enabling more accurate cancer detection. In this context, our paper explores deep learning architectures, namely Artificial Neural Networks (ANN), VGG-19, and AlexNet, for early OSCC detection using histopathology images. The ANN model achieved an accuracy of 89.51 %, while AlexNet and VGG-19 demonstrated superior performance with accuracies of$\mathbf{9 1. 4 8 \%}$and$\mathbf{9 2. 7 9 \%}$, respectively. These findings underscore the potential of deep learning models to enhance OSCC diagnosis, aiding in earlier detection and more effective treatment.