Oral Cancer Analysis for Early Detection using Deep Learning

• Published in: 2024 International Conference on Cognitive Robotics and Intelligent Systems (ICC - ROBINS) pp. 317 - 321
• Main Authors: Shaheer, K M, Edwin, E. Bijolin, Kirubakaran, Stewart, Mathu, T., Ebenezer, V., Thanka, M. Roshni
• Format: Conference Proceeding
• Language: English
• Published: IEEE 17-04-2024
• Subjects: accuracy; AUC-ROC; Convolutional neural network; Convolutional neural networks; Deep learning; diagnosis; Materials reliability; medical image analysis; Oral cancer; Public healthcare; Reliability engineering; Sensitivity; specificity; Training
• DOI: 10.1109/ICC-ROBINS60238.2024.10533923

Worldwide, oral cancer ranks high in terms of death and disability, making it a major concern in public health. In order to effectively treat this condition, a prompt and accurate diagnosis is crucial. Deep learning algorithms used to the analysis of medical images have shown encouraging outcomes in recent years. Several clinical sources provided the pictures used to train and evaluate EfficientNet and CNNs for oral cancer. The goal of training the CNNs was to automatically detect and categorize various stages and forms of oral cancer from images. The EfficientNet method for designing and scaling convolutional neural networks reliably scales the depth, breadth, and resolution dimensions using a compound coefficient. The success of the CNN and EfficientNet models was assessed using standard assessment criteria, including AUC-ROC, sensitivity, specificity, and accuracy. The findings demonstrated that the EfficientNet model is very sensitive and accurate in identifying and categorizing oral cancer, suggesting that it might be a valuable resource for accurate cancer diagnosis.