Title: B-mode breast ultrasound image segmentation techniques: an investigation and comparative analysis

Authors: Madan Lal; Lakhwinder Kaur; Savita Gupta

Addresses: Department of Computer Engineering, Punjabi University, 147002, Patiala, India ' Department of Computer Engineering, Punjabi University, 147002, Patiala, India ' Department of Computer Science and Engineering, University Institute of Engineering and Technology, Panjab University, 160016, Chandigarh, India

Abstract: Breast cancer is the second leading reason for death among women. A commonly used method for detection of breast cancer is ultrasound imaging. Ultrasonic imaging is a low cost, easy to use, non-invasive and portable process, but it suffers from acoustic interferences (speckle noise) and other artefacts. As a result, it becomes difficult for the experts to directly identify the exact shapes of abnormalities in these images. Numerous techniques have been proposed by different researchers for visual enhancement and for segmentation of lesion regions in breast ultrasound images. In this work, different automatic and semi-automatic breast ultrasound image segmentation techniques have been reviewed with a brief explanation of their different technological aspects. Performance of selected methods has been evaluated on a database of 45 B-mode breast ultrasound images containing benign and malignant tumours (25 benign and 20 malignant). For performance analysis of the segmentation methods, results are taken in terms of area and boundary error based quantitative metrics using manually delineated images (by an expert Radiologist) as ground truth/ reference images.

Keywords: B-mode breast ultrasound image; breast tumour; speckle noise; image segmentation; thresholding; region growing; fuzzy clustering; watershed; neural networks; active contour; level set.

DOI: 10.1504/IJCSYSE.2018.091401

International Journal of Computational Systems Engineering, 2018 Vol.4 No.2/3, pp.171 - 184

Received: 26 Oct 2016
Accepted: 20 May 2017
Published online: 30 Apr 2018 *

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