Design of feature extraction and classification of image

Abstract

Medical informatics is a dynamic and transformative field at the intersection of healthcare and technology, where digital data and cutting-edge algorithms are harnessed to enhance the diagnosis and treatment of medical conditions. A crucial component of this domain is the analysis of medical images, which holds immense potential for improving patient care and outcomes. In particular, the classification of skin lesions through image analysis has gained prominence due to the growing prevalence of skin cancer worldwide. The importance of feature extraction and classification in medical informatics, especially for skin lesion image classification, cannot be overstated. newlineSkin lesions, which can be indicative of various dermatological conditions, are often visually complex, with subtle variations in colour, texture, and shape. Manual examination of these lesions can be time-consuming and may not always yield accurate results. Herein lies the significance of feature extraction and classification. Feature extraction involves the identification and extraction of meaningful information from images, which serves as the basis for subsequent classification. By automating this process, it is possible to achieve early and accurate diagnosis, critical for prompt and effective treatment. newlineIn our final contribution, we introduce an innovative cluster-based fusion method that revolutionizes skin lesion image classification. Our two original feature fusion strategies, KFS-MPA (utilizing K-means) and DFS-MPA (employing DBSCAN), harness the power of optimized clustering- based deep feature fusion, complemented by the marine predator algorithm (MPA). We evaluate ten fused feature sets using three different classifiers on both datasets and compare their effectiveness in reducing dimensionality and enhancing accuracy. newlineBy addressing these motivations, the research aims to have a significant impact on patient care, especially in the context of skin cancer diagnosis, and to make a valuable contribution to the broader field of medical infor