Oral Cancer Detection using Metaheuristics

Oral cancer remains a significant global health concern, particularly in regions like India, where it ranks as the sixth most common malignancy, causing approximately 130,000 deaths annually. Traditional diagnostic methods face limitations in precision, particularly in distinguishing malignant cells. This study leverages advancements in deep learning to improve diagnostic accuracy using clinical image datasets. By integrating pre-trained models like VGG19, ResNet50, and EfficientNet, optimized through Manta Ray Foraging Optimization (MRFO), an ensemble deep learning framework is proposed. The method achieves exceptional accuracy (97.4%), sensitivity (95.63%), and specificity (94.12%), showcasing the potential of clinical imaging for accessible and efficient oral cancer diagnosis. Future work includes multiclass cancer diagnosis and evolutionary techniques for optimizing deep learning networks.