A Melanoma Skin Cancer Diagnosis Using Hybrid Feature-Optimized Msvm Classification Model On Dermotoscopic Images

Abstract

With the current advancements in the medical field, skin cancer is measured as a simple infection in the human body. Though the existence of melanoma disease is shown as a form of cancer, it is limitations in classifying it. If Melanoma disease and some other skin lesions are verified in the initial phase signs and symptoms, prediction can be effectively attained to treat them. This dermoscopic skin image plays a significant role in diagnosing a type of skin disease precisely and rapidly. The use of the proposed method is to enhance the SCD (skin cancer detection)SN, SP, Acc rate in dermoscopic images. The research article defines an enhanced plan to detect three skin cancer image categories in early phases. The mentioned input is anSC image which, by using the research technique, the planned system would be classified into cancer or normal categories of images. The clustering method has introduced the segmentation process to divide homogeneous image edges. The image preprocessing steps are done using different steps, such as the filter method, to improve the image attributes. At the same time, the other feature sets are assessed by implementing the RGB color model. GLCM and KPCA feature extraction methods altogether. For classification, MSVM is trained using the Hybrid-featured-optimized MSVM method. Several feature sets are precisely calculated to attain a better outcome using the skin cancer dermoscopic image database HAM10000. The novel work advises that hybrid-featured-optimize MSVM best compared with the other methods, efficiently predicts SC and creates an acc.  rate of 98.0 percent. The outcomes are extremely precisely compared to other methods in a similar field.