Digital Image Based Segmentation and Classification of Tongue Cancer Using CNN

Due to the change in life style after covid-19 there is demand for non-invasive contact less healthcare monitoring systems. In India oral cancer rate are increasing and becoming the community health issue with high mortality rate. Mortality rate can be reduced by identification of disease at initial stages. The major hindrance in disease identification is availability of dedicated hardware at remote and identification at initial stage. Research is going on to make the device portable and less costly using digital images. Also, research is going on to have hybrid algorithm which can segment smaller area abnormal area and classify reliably. This paper focuses on tongue cancer which is one of the types of oral cancer and presents comparison of hybrid algorithm using firefly and watershed transformation to segment smaller area using digital images which reduces cost of dedicated hardware required. 150 digital images are used which are available on internet or provided by cancer hospital for analysis and classification using CNN along with augmentation. 90.48% accuracy is achieved and desirable results are obtained using hybrid algorithm being used.