RWHO: A hybrid of CNN architecture and optimization algorithm to predict basal cell carcinoma skin cancer in dermoscopic images

Abstract

Basal cell carcinoma (BCC) is a type of skin cancer that initiates from the epithelial cells of our skin. Compared to other forms of cancer, BCC infrequently spreads to other parts of the body. It has a risk of local attack and demolition of surrounding tissues. Typically, BCC shows as one or numerous small, glowing nodules exhibiting central depressions. These knots are commonly found on the sun-exposed skin areas of older adults. Many dermatoscopic methods are available for diagnosing and predicting such kinds of skin cancers. But, medical professionals find it difficult to diagnose at some kind of images at the early stages. An automated methodology to predict such types of skin lesions would be better for such a diagnosis. In the present work, a new computer-assisted algorithm called RESNET50-WHO (RWHO) has been introduced to predict and diagnose BCC skin cancer. The method uses a combination of deep learning algorithm RESNET 50 and a metaheuristic algorithm, called wildebeest herd optimization (WHO) Algorithm to do prediction. The initial features from the images are extracted using RESNET 50. The output is given to the WHO algorithm to extract the beneficial features to reduce the time complexity. The method is tested using the PH2 dataset. The results obtained using the proposed algorithm is compared with the state-of-art optimization algorithms and evaluated. The conclusive findings specify that the proposed algorithm beats the comparative methods, yielding superior results