Deep Learning for Oral Cancer: Enhanced Regression Based Convolution Neural Network with Dropout Technique

Background and Aim: Deep learning techniques have been proven as useful tools in detecting and diagnosing oral cancer. This early and accurate prediction depends on the accurate classification of oral cancer images through an automated and computer-aided system. Convolution neural network technique has been succeed in classifying the images and predicting oral cancer. Methodology: The proposed system aims to increase the accuracy of classification and decrease the processing time. The proposed system comprises of regression based convolution neural network with dropout technique in the fully connected layer of neural network to decrease the over-fitting error which summarizes the information. Results: Experimental results show that the proposed solution has improved the accuracy of classification by approximately 3% and reduced the processing time by almost 0.25–0.30 s on average. Conclusion: The proposed system seems to be based on the accurate classification of oral cancer cells in three main classes as normal, benign and malignant cells. It enhances the classification accuracy and reduces the processing time.