Tuberculosis mycobacterium segmentation using deeply connected membership tweaked fuzzy segmentation network

作     者：Shiny, A. Amala Sivagami, B. 

作者机构：Department of Computer Applications S.T. Hindu College Nagercoil India Affiliated to Manonmaniam Sundaranar University Tirunelveli India Department of Computer Science & Applications S.T. Hindu College Nagercoil India 

出 版 物：《Multimedia Tools and Applications》 (Multimedia Tools Appl)

年 卷 期：2025年第84卷第10期

页      面：6899-6929页

核心收录：

学科分类：0831[工学-生物医学工程（可授工学、理学、医学学位）] 1007[医学-药学(可授医学、理学学位)] 100706[医学-药理学] 1002[医学-临床医学] 070207[理学-光学] 1001[医学-基础医学(可授医学、理学学位)] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 081202[工学-计算机软件与理论] 0702[理学-物理学] 

主　　题：Image segmentation 

摘      要：Tuberculosis (TB) is a contagious disease that spreads through the air when an infected person coughs, sneezes, or talks. TB is a bacterial infection caused by Mycobacterium tuberculosis (MTB). Nowadays, TB diagnosis is mainly influenced by the segmentation of MTB objects in Ziehl-Neelsen (ZN)-stained microscopy images. There are a number of segmentation algorithms that have been extensively examined in the literature, and these are still prone to issues like less accuracy due to over and under segmentation as well as unclear edges in bacterium objects. Hence, this paper presents a novel MTB segmentation method entitled ‘TB Mycobacterium Segmentation using Deeply connected Membership tweaked Fuzzy Segmentation Network (TBMS-DMFSN)’. This method takes the ZN-stained MTB color microscopic image as input and generates the segmented result. The own contribution of this paper is the FCM-based Mycobacterium segmentation algorithm called ‘Deeply connected Membership modified Fuzzy Segmentation Network (DMFSN),’ which is a deeply connected fuzzy network. The DMFSN network uses a back-propagated fuzzy network to separate the Mycobacterium objects. The proposed TBMS-DMFSN method incorporates eight optimum feature images from four different color spaces, lightness enhancement, and Gamma correction algorithms. Experimental evaluation is constructed with the help of both online and real-time clinical databases. The proposed TBMS-DMFSN approach has overall average segmentation accuracy (SA) of 95.26%, whereas the second-best method has a segmentation accuracy of 93.03%. As a result, the proposed approach raises the SA values by up to 2.226%. The achieved results reflect that the proposed method produces higher accuracy compared to state-of-the-art methods. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.