Type-reduced vague possibilistic fuzzy clustering for medical images. (April 2021)

• Record Type: Journal Article
• Authors: Bose, Ankita; Mali, Kalyani
• Is Part Of: Pattern recognition. Volume 112(2021)
• Publication Date: 2021-04
• Subjects: Fuzzy membership -- Typicality -- Vague set -- Type-reduction -- Medical images; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2020.107784
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: Effective hybridization of fuzzy and possibilistic membership to obtain an optimum solution. A vague environment is created using this hybridization. It takes probabilistic membership as the upper bound and possibilistic membership as the lower bound. Experimental results show the efficiency of the proposed work on the diagnosis of medical images. The novelty of the proposed approach is mentioned in terms of accuracy and error detection. Abstract: Soft computing provides the framework for dealing with the uncertainty and imprecision inherent in real-life applications. Soft computing has become a long-standing notable paradigm for medical image processing. A typical fuzzy clustering uses the fuzzy membership function. Nevertheless, there is an alternative membership representation, known as typicality or possibilistic membership. Unlike fuzzy membership that is probabilistic in nature, typicality represents an absolute membership and it is the degree of belonging of an object to a class that does not depend on its distances from the other classes. However, both fuzzy membership and typicality play important role in assigning membership to an object. This study proposes a novel clustering model that creates a vague environment enriched with the concept of fuzzy membership and typicality, while the use of type-reduction plays an essential role in capturing all the vagueness present in the data set. The proposed model is called type-reduced vague possibilistic fuzzy clustering (TVPFC), and we use MRI images to demonstrate its superior robustness over that of FCM (fuzzy c-means), PCM (possibilistic c-means), VCM (vague c-means) and IPFCM (interval-valued possibilistic fuzzy c-means).