Analysis of clustering approaches for skin lesion segmentation using dermoscopic images Thamizhamuthu R1, Manjula D2 1Research Scholar, Department of Computer Science and Engineering, CEG Campus, Anna University, Chennai 2Professor and Head of the Department Department of Computer Science and Engineering, CEG Campus, Anna University, Chennai Online published on 1 November, 2018. Abstract In this paper, clustering approaches are analyzed for skin lesion segmentation using dermoscopic images. Three widely used machine learning approaches for image segmentation are K-Means Clustering (KMC). Fuzzy C-Means Clustering (FCMC), and Expectation-Maximization (EM) algorithm. The difference between KMC and FCMC lies in the partitioning method. The former one uses hard partitioning, and the later uses a variable degree of membership. In the EM algorithm, statistical methods are employed for distance calculation whereas, in KMC, the Euclidean distance measure is used. The segmentation results of individual clustering approaches are combined to get the refined skin lesion. Results show that the combined segmentation provides promising results for skin lesion segmentation in comparison with KMC, FCMC and EM algorithm. Top Keywords Skin lesion segmentation, clustering techniques, k-means clustering, fuzzy clustering, expectationmaximization algorithm. Top |