An Approach Towards the Detection of Ocular Cancer using Fast R-CNN Technique

Although eye melanoma is a rare condition, it is the most frequent form of cancer, according to malignancy. Like other malignancies, eye melanoma can be treated in the majority of instances with the right diagnosis, although this is the most difficult and problematic part of the process. This study describes a convolutional neural network-based automated technique for detecting eye melanoma (CNN). A standard database is used to provide 170 pre-diagnosed samples, which are then processed to create a sample of lower resolution sent to the Convolution Neural Network (CNN) architecture. In order to diagnose eye melanoma, the suggested approach does away with independent feature extraction and classification. Despite requiring a significant amount of computing, the suggested method outperforms artificial neural network (ANN) eye melanoma detection with an accuracy rate of 91.76%. This study presents a deep learning architecture for CNN-based automated eye melanoma detection. CNN is made up of one or more convolution layers, subsampling layers, one or more layers that are fully linked (like a regular ANN), and finally one or more convolution layers. The fundamental benefit of adopting Convolution Neural Network (CNN) is that, its training is easier and has fewer parameters than artificial neural network with the identical number of layers hidden.