Affiliations: [a] Scientific Research Group in Egypt (SRGE), Egypt | [b] Artificial Perception for Intelligent Systems and Robotics (AP4ISR), Institute of Systems and Robotics (ISR), University of Coimbra, Coimbra, Portugal | [c] Ingeniarius, Lda., Mealhada, Portugal | [d] Faculty of Computers and Information, Cairo University, Cairo, Egypt | [e] Department of ECE, Karunya University, Coimbatore, India
Correspondence:
[*]
Corresponding author: D. Jude Hemanth, Department of ECE, Karunya University, Coimbatore, India. Tel.: +91 9443001874; E-mail:jude_hemanth@rediffmail.com
Abstract: This paper presents a Fuzzy C-Means based image
segmentation approach that benefits from the Minkowski distance as the
dissimilarity measure, denoted as FCM-M, instead of the traditional
Euclidean distance, herein identified as FCM-E. The proposed approach was
applied on Liver CT images, and a thorough comparison between both methods
was carried out. FCM-M provided better accuracy when compared to the
traditional FCM-E, with an area under the ROC curve of 85.44% and
47.96%, respectively. In terms of statistical significant analysis, a
twofold benefit was obtained from using the proposed approach: the
performance of the image segmentation procedure was maintained, or even
slightly increased in some situations, while the CPU processing time was
significantly decreased. The advantages inherent to the proposed FCM-M pave
the way to a whole new chain of fully automatic segmentation methods.
