Title: Particle swarm optimisation K-means clustering segmentation of foetus ultrasound image

Authors: Deepa Parasar; Vijay R. Rathod

Addresses: Research Scholar, Faculty of Engineering, Pacific Academy of Higher Education and Research University, Udaipur, Rajasthan, India ' Department of Electronics and Telecommunication, St. Xavier's Technical Institute, Mahim, Mumbai, Maharashtra, India

Abstract: The purpose of medical image segmentation is to extract information such as volume, shape, motion of organs for detecting abnormalities from the medical image for improvement and fast diagnosis. In this paper, a segmentation algorithm has been implemented for foetus ultrasound image by particle swarm optimisation (PSO) K-means clustering algorithm with fuzzy filter. Impulsive noise inherent in ultrasound image has been removed using fuzzy filter. Then, PSO K-means clustering segmentation method is applied for partitioning foetus ultrasonic images into multiple segments, which applies an optimal suppression factor for the perfect clustering in the specified data set. Experimental results show that the proposed algorithm outperforms other segmentation algorithms like seeded region growing using PSO, fuzzy C-means and watershed in terms of segmentation accuracy for speckle noise added to foetus ultrasound medical images.

Keywords: fuzzy C-mean segmentation; fuzzy filter; K-means segmentation; PSO; seeded region segmentation; ultrasound imaging; watershed segmentation.

DOI: 10.1504/IJSISE.2017.084569

International Journal of Signal and Imaging Systems Engineering, 2017 Vol.10 No.1/2, pp.95 - 103

Received: 25 Jul 2016
Accepted: 03 Feb 2017
Published online: 14 Jun 2017 *

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