Deep Long and Short Term Memory with Tunicate Swarm Algorithm for Skin Disease Detection and Classification

Abstract

The development and implementation of cost-effective and efficient screening technologies is important. To address these concerns, we have introduced a unique method to detect skin diseases. Each photo is first pre-processed and cropped to pixel size. Six square fields are used to split these pictures into pixels. Techniques for enlarging images, such as rotation, mirroring, and enhancement, are employed to minimize the quantity of parameters needed for further processes. An kernel-weighted fuzzy local information or the C-means clustering model (K-FCM) is used to properly segment cancer-affected regions. Texture and colour features are then extracted. Finally, a deep long-term and short-term memory (DLTM)-based tunicate group algorithm (TSA) is used to detect skin diseases and classify both normal and abnormal classes. The experiment was carried out using MATLAB, and photos were gathered from the Helllev University Hospital in Denmark. According to the comparative analysis results, the proposed DLSTM-TSA outperforms competing products in terms of F-score, sensitivity, and precision.