Consignment of Epidermoid Carcinoma using BPN, RBFN and Chebyshev Neural Network

Oral cancer is the most dangerous and fast-spreading cancer in the world with the least survival rate. Oral cancer appears in and around the mouth. Of the oral epithelial carcinoma, cancer that occurs in the lip has the lowest mortality rate and tongue cancer has the highest mortality rate. Early-stage identification and treatment of Oral Squamous Cell Carcinoma is the most important aspect to avoid or reduce the rate of mortality. For early detection, other than physical examinations, various imaging techniques are available such as histopathological imaging, biomedical imaging, Ultrasound, MRI, CT, and Confocal scanning techniques. Multiclass classification is a major role in a real-world application. Tumor, Node and Metastasis are the most important aspect in the staging of oral cancer. Artificial Neural Networks and Machine Learning Techniques work much more effectively to provide better accuracy in medical image classification. For the classification of Oral Squamous Cell Carcinoma images and non-cancerous images, it is tested with three different networks viz Backpropagation network (BPN), Radial Basis Function network (RBF) and Chebyshev neural network. Dimensionality reduction has been done using Principal Component Analysis and explained using a variance plot. The r-square coefficient, which gauges how well the data fit the regression line, shows that, when compared to raw data, the PCA components boost the data's dependability. Of these BPN works well in the Epithelial carcinoma histopathological image classification when compared with RBF and Chebyshev network.