Deep Learning for Oral Cancer Detection: AFF-U-Net Performance Analysis

OSCC is a leading cause of cancer-related fatalities around world, with early detection being critical for improving outcomes. This study proposes a deep learning method approach using the AFF-U-Net architecture, an enhanced U-Net model with advanced attention mechanisms and feature fusion techniques, to detect and classify OSCC in histopathology images. To mitigate overfitting and class imbalance, preprocessing techniques like normalization and data augmentation were employed, boosting the model's generalization ability. The AFF-U-Net achieved an impressive test accuracy of 93.75%, demonstrating strong performance in terms of precision, recall, and F1-score metrics. Confusion matrix analysis confirmed its reliability in differentiating cancerous and healthy tissues. These findings demonstrate AFF-U-Net's potential as a powerful tool for automated OSCC detection, improving diagnostic accuracy and effectiveness, and laying the foundation for practical implementation in real-world scenario clinical adoption to enhance patient care.