Oral Cancer Detection Using Convolutional Neural Networks

In developing nation oral cancer makes the significant threat to human life. The existing system uses deep learning technique for detection of oral cancer in medical imaging. The two powerful tools of deep learning techniques are Convolutional Neural Networks and Deep Belief Network [1]. PSOBER- Particle Swarm Optimization and AI-Biruni Earth Radius a hybrid optimization algorithm is used to optimize the CNN and DBN. To overcome the outperformance of the existing method standard dataset of biomedical images are used for demonstration of the promising results. To test and validate the significance and stability one-way ANOVA and Wilcoxon signed-rank is used. For screening oral cancer detection in clinical setting this technique is used. However, the accuracy and to find the early detection of oral cancer and to make the less suffering of the patients the further research is needed. The proposed method of oral cancer detection using deep learning technique includes the enhancement of the images. In this method salt and pepper noise detection is used to find the accurate results using segmentation of the cells for thresholding. Back Propagation Neural Networks (BPNN) is used for the classification of the oral cancer. Tumor partition and features extraction are done in the proposed method. Feeding the loss backward through neural networks layers to fine tune the weights and the BPNN involves the error rate of forward propagation. Back propagation neural networks contain two signals. They are Error signal and Back signal.