Analysis of basic neural network types for automated skin cancer classification using Firefly optimization method (Retracted Article)

In recent days, cancer is a deadly disease because of its spreading nature to other cells, and this disease is not identified at an early detection stage. Generally, the cancer is detected with the help of a biopsy method, which is a painful approach. Due to the development of technology, nowadays, it is identified with the help of image processing methods. Here, the image processing approach is used for identifying and classifying the skin cancer types, namely melanoma, common and atypical nevi. The methods used earlier for the detection and classification are artificial skin leison merging, Raman spectroscopy and back-propagation networks. Cancer is classified into many types like blood cancer, bone, colon, and stomach and skin cancer. Among these cancer types, skin cancer can be a dreadful disease, which is detected and then treated at the starting stage of the disease. Hence, this paper proposed an optimized neural and fuzzy approach for skin cancer classification. The fuzzy c-means segmentation is used for the detection of the cancer region. Firefly optimization determines the dominant feature for the training of the neural network. The dominant feature is determined by reducing the error rate of the classifier. The overall process is evaluated with the help of evaluation metrics like accuracy, specificity and sensitivity. In this proposed method, the best result is achieved for the pattern net by improving its accuracy by 4.9% from its previous Moth-Flame Optimization based classification in its evaluation.