ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

Fuzzy clustering of Acute Lymphoblastic Leukemia images assisted by Eagle strategy and morphological reconstruction

Das, A and Namtirtha, A and Dutta, A (2022) Fuzzy clustering of Acute Lymphoblastic Leukemia images assisted by Eagle strategy and morphological reconstruction. In: Knowledge-Based Systems, 239 .

[img] PDF
kno_bas_sys_239_2022.pdf - Published Version
Restricted to Registered users only

Download (4MB) | Request a copy
Official URL: https://doi.org/10.1016/j.knosys.2021.108008


Patients with Acute Lymphoblastic Leukemia (ALL) require prompt diagnosis since it has the chance to become fatal if neglected for a few weeks. The microscopic image of lymphocyte cells creates doubts even for expert pathologists because normal lymphocytes cells and ALL blast cells are both very smooth. Therefore, proper segmentation of White Blood Cells (WBC) is a very crucial aspect of the task. Consequently, the focus of the study is on segmenting the WBCs in Acute Lymphoblastic Leukemia (ALL) images utilizing classical crisp and fuzzy clustering approaches like K-means (KM) and Fuzzy C-means (FCM). But these classical clustering approaches are very sensitive to noise and initial cluster center initialization and hence trapped in local optima. As a result, these techniques may produce incorrect cluster centers. Firstly, researchers are employing Nature-Inspired Optimization Algorithms (NIOAs) as an alternate methodology for both crisp and fuzzy clustering problems to solve the initial cluster center initialization issue. Therefore, using a two-stage Eagle Strategy based on Stochastic Fractal Search (SFS) method, this research proposes a fuzzy clustering methodology. Secondly, morphological reconstruction has been employed for filtering the membership matrix to guarantee noise-immunity. A scrupulous parallel study is performed among the proposed eagle strategy based fuzzy clustering depending on morphological reconstruction with some well-known NIOA based fuzzy clustering and crisp clustering approaches, and classical clustering methodologies like KM and FCM in view of a collection of color ALL images and regular performance metrics. Experimental results demonstrate that recommended ES-SFS based fuzzy clustering technique with morphological reconstruction surpasses the most of utilized approaches in words of computation effort, quality metrics, and robustness. Additionally, to get rid of the random effect in the achieved numerical results, a non-parametric strategy is used for statistical validation. © 2021 Elsevier B.V.

Item Type: Journal Article
Publication: Knowledge-Based Systems
Publisher: Elsevier B.V.
Additional Information: The copyright for this article belongs to Elsevier B.V.
Keywords: Blood; Chromosomes; Cluster analysis; Cytology; Diagnosis; Diseases; Fuzzy clustering; Image reconstruction; K-means clustering; Random processes; Stochastic systems; Swarm intelligence, Acute lymphoblastic leukaemias; Cell-be; Cell/B.E; Cell/BE; Clustering approach; Images segmentations; K-means; Morphological reconstruction; Optimisations; Pathology image, Image segmentation
Department/Centre: Division of Interdisciplinary Sciences > Computational and Data Sciences
Date Deposited: 20 Jan 2022 06:35
Last Modified: 20 Jan 2022 06:35
URI: http://eprints.iisc.ac.in/id/eprint/70965

Actions (login required)

View Item View Item