Melanoma is one among the first cancer which effects on skin and is the deadliest diseases which is affected after breast cancer. The color images of the skin is taken, there is comparatively a high similarity appears between different skin lesion such as melanoma and nevus, which wax the effort of the detection and diagnosis. A capable robust system is necessary to detect skin lesion and the process classification is essential for early detection and prognosis to save human life with minimum effort and time. In this research paper, develop an improved robust technique for the classifying melanocytic tumors either as malignant or benign on consideration with the digital dermoscopy image’s texture features. The concept proposed here follows in three steps: firstly, The skin lesion localization for the image is measured using an adaptively regularized kernel-based fuzzy 𝐶-means clustering (ARKFCM); secondly, the features descriptive of tumor texture are extracted using GLCM, MSD, LTCOP & LCP and finally, the lesion objects are classified using a classifier based on a fuzzy based hybrid kernel SVM model. The overall performance of the proposed concept is verified and evaluated by the metrics: accuracy, sensitivity and specificity where the obtained values of these measures are 92.58%, 93.90%, and 90.66% respectively. The performance of the proposed system is compared with the performance Adaboost and Random Forest (RF) classification methods where it is seen that the classification rate of the proposed method overtook the performance of the previous classification methods.
Volume 11 | 06-Special Issue
Pages: 1578-1593