A neutrosophic-entropy based clustering algorithm (NEBCA) with HSV color system: A special application in segmentation of Parkinson’s disease (PD) MR images

• Published in: Computer methods and programs in biomedicine Vol. 189; p. 105317
• Main Author: Singh, Pritpal
• Format: Journal Article
• Language: English
• Published: Ireland Elsevier B.V 01.06.2020
• Subjects: Algorithms; Color; Entropy; Humans; Image Interpretation, Computer-Assisted - statistics & numerical data; Image segmentation; Magnetic resonance (MR) images; Magnetic Resonance Imaging; Neuroimaging; Neutrosophic set (NS) theory; Parkinson Disease - diagnostic imaging; Parkinson’s disease (PD); Visualization; Magnetic resonance (MR) images; Image segmentation; Visualization; Parkinson’s disease (PD); Neutrosophic set (NS) theory
• ISSN: 0169-2607; 1872-7565; 1872-7565
• DOI: 10.1016/j.cmpb.2020.105317

•This article presents neutrosophic set theory based segmentation method.•The method uses HSV system for segmented regions visualization.•Application is shown in MR images of Parkinson’s disease.•The method exhibits high accuracy rate. Background and objectives: Brain MR images consist of three major regions: gray matter, white matter and cerebrospinal fluid. Medical experts make decisions on different serious diseases by evaluating the developments in these areas. One of the significant approaches used in analyzing the MR images were segmenting the regions. However, their segmentation suffers from two major problems as: (a) the boundaries of their gray matter and white matter regions are ambiguous in nature, and (b) their regions are formed with unclear inhomogeneous gray structures. For these reasons, diagnosis of critical diseases is often very difficult. Methods: This study presented a new method for MR image segmentation, which consisted of two main parts as: (a) neutrosophic-entropy based clustering algorithm (NEBCA), and (b) HSV color system. The NEBCA’s role in this study was to perform segmentation of MR regions, while HSV color system was used to provide better visual representation of features in segmented regions. Results: Application of the proposed method was demonstrated in 30 different MR images of Parkinson’s disease (PD). Experimental results were presented individually for the NEBCA and HSV color system. The performance of the proposed method was evaluated in terms of statistical metrics used in an image segmentation domain. Experimental results, including statistical analysis reflected the efficiency of the proposed method over the existing well-known image segmentation methods available in literature. For the proposed method and existing methods, the average CPU time (in nanosecond) was computed and it was found that the proposed method consumed less time to segment MR images. Conclusion: The proposed method can effectively segment different regions of MR images and can very clearly represent those segmented regions.