DEEPORCD: Detection of Oral Cancer using Deep Learning

Abstract

Abstract Oral cancer is a widespread and complex cancer with a high severity. Oral cancer is the eighth most common cancer in the world in India, with 130,000 deaths in each year. The tumor occurs in the salivary glands, tonsils, as well as in the neck, face and mouth. There are various diagnostic methods for oral cancer, such as a biopsy, in which a small tissue sample is taken from a part of the body and tested under a microscope also some screening methods. But the downside is that cannot clearly identify cancer cells and cannot classify the number of cells affected by cancer, so in this work cancer cells will find and classify that affected in the oral area through digital processing technology. The use of advanced technologies and an in-depth learning algorithm are possible for early detection and classification. This work uses three characteristics-extracting techniques such as the bag histogram of oriented gradients, wavelet features and the Zernike Moment. Once retrieving the texture characteristics, the fuzzy particle swarm optimization algorithm (FPSO) is applied to choose the best characteristic. Finally, these characteristics were classified using the Convolution Neural Network (CNN) classifier. For comparison of the efficiency of the proposed method, Recall Rate, Classification Accuracy, Precision Rate, and Error Rate. Evaluation outcomes demonstrated that the combination of ABC, FPSO and CNN performs better in the detection of oral cancer.