Oral Cancer Detection Using Deep Learning

• Published in: 2024 International Conference on Science Technology Engineering and Management (ICSTEM) pp. 1 - 6
• Main Authors: Chavva, Renuka, S, Jeba Priya, Mathu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 26.04.2024
• Subjects: Accuracy; Analytical models; Cancer detection; classification; CNN; Data models; Deep learning; Deep learning image analysis; DenseNet; early detection; Oral cancer; Tongue; Transfer learning
• DOI: 10.1109/ICSTEM61137.2024.10561172

Use of deep learning methods, and more especially the DenseNet architecture, for the purpose of oral cancer detection is the primary goal of this research. Oral cancer, thrush, lichens, hairy tongue, leukoplakia, and healthy tongue photos are all part of the oral conditions dataset. The dataset includes oral images from both healthy individuals and those with illness. We added new classification layers after fine-tuning a pre-trained DenseNet169 model that had been trained on ImageNet using transfer learning. Many methods of data augmentation were used to improve model resilience. A performance comparison using the LeNet model was done. The DenseNet based model demonstrated exceptional performance, attaining 94.08% accuracy, 94.16% precision, 94.7% recall, and 94.7% F1 score. By comparison, the LeNet model performed less, with an F1 score of 63.01%, recall of 64.03%, accuracy of 64.02%, and precision of 64.06%. Medical image analysis relies on data preparation, model selection, and assessment methods this study shows that deep learning might be useful for detecting oral cancer. Improving the model's performance for real-world deployment in oral cancer detection might be achieved by more tuning and testing.