Detection of Skin Cancer Using SFLA Based Apex Component Analysis Network

Diagnosing skin lesions using medical image analysis is still difficult. There are many different types of skin cancer, but skin lesions are quite prevalent. the most cutting-edge tools for identifying skin cancer is dermoscopy. However, a high number of training samples is often required for Deep learning-based approaches. When there are few samples available, it is challenging for a deep learning model to produce a representative representation of features for detecting skin lesions. In this work, we present an optimised simple deep learning model (OSDL) that can improve accuracy even when there is a scarcity of training data. In OSDL, the network is built using samples of skin cancer images and their innate features. The resultant model would then be able to create more robust feature expression with fewer data points. The suggested model, dubbed SFLA-OSDL, employs the Shuffled Frog Leaping Algorithm (SFLA) to fine-tune OSDL's hyper-parameters. For the purpose of deep feature extraction from skin cancer pictures, we have developed a novel network we call the Apex Component Analysis Network (ACAN). On the benchmark dataset ISBI-2016, the provided technique is exposed to outperform the high-tech tactics in relationships of accuracy and speed.