Kapur’s Entropy based Hybridised WCMFO Algorithm for Brain MR Image Segmentation

Abstract

This article proposes a Kapur-based hybridized Water Cycle and Moth-Flame Optimisation (WCMFO) algorithm that combines a water cycle algorithm (WCA) and moth flame optimisation (MFO) in multilevel thresholding of brain MR image segmentation. The WCMFO algorithm, proposed by Khalilpourazari and Khalilpourazary, gives both WCA and MFO advantages, while avoiding some of the drawbacks of either approach on its own, as demonstrated by faster convergence with broader exploration and exploitation capabilities. Experiments on 10 axial, T₂-weighted test images were performed using Kapur entropy as the objective function at a threshold level of m = 2–5. The spiral movement of the behaviour of the moths is used for better exploitation in the WCA to find the global optimum values. WCMFO results, such as objective function value, peak signal to noise ratio, Central processing unit time and standard deviation, are collated and compared with other existing adaptive wind-driven optimization algorithm, adaptive bacterial foraging and particle swarm optimization algorithms. Experimental findings and comparison demonstrated that hybridized WCMFO algorithm was superior to the other algorithms. Moreover, the best segmentation is achieved on grey matter, white matter and cerebrospinal fluid that allows for better clinical decision-making and diagnosis in the medical field. Therefore, the proposed multilevel thresholding-based hybridized WCMFO algorithm is believed to be the most prominent preference for segmenting such complex brain images.

KEYWORDS:

• Brain MR image
• Hybrid WCMFO algorithm
• Image segmentation
• Kapur entropy
• Multilevel thresholding

Additional information

Notes on contributors

A. Renugambal

A Renugambal is currently working as a teaching fellow in the Department of Mathematics at the University College of Engineering Kancheepuram (A constituent College of Anna University, Chennai), Kancheepuram, Tamil Nadu, India. She is a research scholar in Anna University, Chennai. In the year 2006, she completed MPhil in Mathematics at Thiruvalluvar University and MSc in Mathematics at Madras University in the year 2004. She has published four research papers in refereed journals. Her research includes applied mathematics, image segmentation and applications of nature inspired algorithms.

K. Selva Bhuvaneswari

K Selva Bhuvaneswari is currently working as an assistant professor in the Department of Computer Science and Engineering at the University College of Engineering Kancheepuram, (A constituent College of Anna University, Chennai), Kanchieepuram. She has received PhD in Information and Communication Engineering from Anna University, Chennai in the year 2015. She has published nine research papers in reputed international and national journals. Her current research includes image processing, semantics and data mining. E.mail: [email protected]