Title: Mass segmentation in mammograms using a kernel-based fuzzy level set method

Authors: J. Anitha; J. Dinesh Peter

Addresses: Department of CSE, Karunya University, Coimbatore, Tamil Nadu, India ' Department of IT, Karunya University, Coimbatore, Tamil Nadu, India

Abstract: In this paper, a Kernel-Based Fuzzy Level Set (KFLS) method is proposed to facilitate the automatic segmentation of lesions in mammogram image. Initially, the mammogram image is pre-processed to remove the noises and to extract the breast profile. Kernel-based Fuzzy C-Means (KFCM) segmentation is applied to segment the pre-processed image into a number of clusters based on the breast structure. The cluster with Region of Interest (ROI) is automatically identified to remove background and normal tissues. Finally, the mass region extracted from the KFCM is used as an initial contour for the level set segmentation to refine the mass boundary. The result of KFCM is also used to automatically initialise the control parameters of level set segmentation algorithm. This methodology is validated on 300 mammogram images of DDSM database with diagnosed mass. The experimental results demonstrate that the proposed method achieves a sensitivity of 93.32% and accuracy of 94.31%.

Keywords: kernel-based fuzzy clustering; level sets; mass segmentation; morphological operations; mammograms; mammography; breast cancer screening; lesion segementation; mammogram images.

DOI: 10.1504/IJBET.2015.072933

International Journal of Biomedical Engineering and Technology, 2015 Vol.19 No.2, pp.133 - 153

Received: 20 Jan 2015
Accepted: 13 Apr 2015
Published online: 08 Nov 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article