Performance Analysis for Quality Measures using K means Clustering and EM Models in Segmentation of Medical Images
R. Hari Kumar1, B. Vinoth Kumar2, G. Karthick3
1Dr. R. Harikumar, Professor, Department of ECE, Bannari amman Institute of Technology, Sathyamangalam, India.
2B. Vinoth kumar, Assistant Professor(Senior Grade), Department of EEE, Bannari amman Institute of Technology, Sathyamangalam, India.
3G. Karthick, PG Scholar, Department of ECE, Bannari Amman Institute of Technology, Sathyamangalam, India.
Manuscript received on November 25, 2011. | Revised Manuscript received on December 14, 2011. | Manuscript published on January 05, 2012. | PP: 74-80 | Volume-1 Issue-6, January 2012. | Retrieval Number: F0271111511/2012©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The main objective of this paper is to compare the performance of quality measures towards the segmentation of medical images using K-means clustering and EM models. Three types of medical images such as MRI, X-rays and Ultrasonic images are studied. The K-means clustering shows that the non intactness of the clusters. As cluster size increases the edges are brittle and compactness of the clusters get altered. Hence expectation maximization models are utilized to segment the images for better edge perseverance and compactness of clusters at larger size. The quality measures like PSNR, average difference, structural content, image fidelity and normalize coefficients are calculated for both methods. The EM models shows one dB increase in PSNR values than the K-means clustering. At less number of clusters AD value of EM models mitigates the compactness of the cluster centers.
Keywords: Segmentation, K-means clustering, EM models, Quality measures.