Hybrid machine learning based model for automatic detection of melanoma

Abstract

The major objective of this thesis is to develop a robust, accurate, and reliable automatic melanoma detection system. Our suggested approach is performed in four phases: In the first phase, the input image is pre-processed. The lesion region is extracted in the second phase, in the third phase, extraction of features is performed, and finally in the fourth phase lesion is classified as benign or melanoma. The present work has been performed to recognize the lesion from publicly available datasets. This thesis is divided into six chapters. A brief summary of the thesis chapters is as follows: Chapter one includes the introduction of the automatic melanoma detection system. The applications of AMDS along with the challenges faced by such a system are also mentioned in this chapter. Chapter two discusses the major research contributions available in the literature. Many researchers have carried out their work in this research area for the past twenty-three years and still, it is an active research area. Techniques and methods employed in the literature are discussed. The literature review covers the research articles and papers published between 2000 and 2023. In Chapter three, two new AMDS algorithms named LePrePro (Lesion Pre-Processing Technique) and LeFet (Lesion Feature Extraction Technique) are proposed. Using the extracted features of lesion and classification techniques the input images of the lesions are used to classify the lesion as benign or melanoma in the proposed approach. newline