Affiliations: [a] School of Electrical & Electronic Engineering, Universiti Sains Malaysia, Engineering Campus, Nibong Tebal, Penang, Malaysia | [b] School of Medical Science, Universiti Sains Malaysia, Health Campus, Kota Bharu, Kelantan, Malaysia
Abstract: Recently, numerous clustering methods have been tested and used in the segmentation process especially for medical imaging applications. The medical imaging analyses require high accuracy of result percentage. According to this issue, numerous studies have been carried out in medical imaging field such as segmentation technique. According to the high expectation of segmentation output, researchers have been attracted to study and develop new methods of segmentation techniques. Certain medical images can be classified as hard to process and understand. Therefore, several clustering algorithms have been proposed to meet the expectation from medical view as well as to produce better segmentation performance for medical images. In this study, the modified moving k-means algorithm is proposed for the segmentation problems. The objective of the modified moving k-means algorithm is to introduce the better technique of finding the nearest center for each pixel to be classified into different clusters. Then, the modified moving k-means algorithm is compared to the fuzzy c-means, k-means and moving k-means algorithms. The comparison result shows that the modified moving k-means algorithm produced better segmentation quality as compared to the other clustering techniques.
