Empowering Oral Squamous Cell Carcinoma Detection With Deep Learning: Insights From Convolutional Neural Network Analysis of Histopathological Images

The global health is heavily impacted by Oral Squamous Cell Carcinoma (OSCC), marked by high mortality rates often due to late diagnoses. This research delves into leveraging Convolutional Neural Networks for the early detection of OSCC. The model's architecture is intricately designed to discern critical features of OSCC, incorporating convolution, max-pooling, and dense layers, culminating in a sigmoid function for binary classification. This research findings reveal the model's high proficiency in OSCC identification, achieving 98.49% training accuracy, 86.89% validation accuracy, and 89.37% testing accuracy. Notably, for the Normal class, the model demonstrated a robust precision of 0.88, recall of 0.90, and the F1-score of 0.89, whereas for the OSCC class, precision is 0.90, recall is 0.88, and the F1-score is 0.89, underscoring its effectiveness in differentiating between normal and cancerous tissues. This study suggests CNNs as a viable and promising approach for OSCC early detection, potentially transforming screening and improving patient outcomes.