Performance analysis of linear layer neural networks for oral cancer classification

A plethora of disorders are found in human oral mucosa. A variety and huge number of lesions and diseases in human oral mucosa have been clinically identified and classified. Most lesions have the possibility to develop into oral cancer. The initial diagnosis of oral cancer is to inspect the ocular regions carefully and register the oral cavity of the patient as true-color digital images. The decision about the further treatment of the oral cancer patient is predominantly depending on the appearance of the lesion. The primary intention of this work is to analyze the clinical features, its respective diagnostic procedures and the required treatment suitable for the oral cancer patients. The oral cancer staging is classified into two types namely pathological and clinical. A lot of prognostic and most efficient tools have been developed in Tumour-Node-Metasis (TNM) stages. The classification accuracy of the TNM staging system is compared to the Linear Layer Neural Network. The study is analyzed for 75 oral cancer patients and the input variables are considered here as the TNM variables. Results show that an average classification accuracy of 100% is obtained in T1 stage, 85.19% is obtained in T2 stage, 84.21% is obtained in T3 stage and 94.12% is obtained in T4 stage.