Variational methods with application to medical image segmentation: A survey
Shu, Xiu Li, Zhihui Chang, Xiaojun Yuan, Di
Shu, Xiu
Li, Zhihui
Chang, Xiaojun
Yuan, Di
Author
Supervisor
Department
Computer Vision
Embargo End Date
Type
Journal article
Date
2025
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
The segmentation model based on the variational method is gradually applied to medical image segmentation, but medical image segmentation still faces many challenges, such as the interference of intensity inhomogeneity, noise and unclear boundaries, etc. To solve the difficult challenge of image segmentation, various variational methods have appeared in the past 20 years. Based on the variational method, there are many methods, which are divided into the region-based model, the edge-based model, introducing other algorithms and adding the constraint term, etc. Each of these methods has its advantages and disadvantages, but there is still a certain distance from being used in clinical medicine. Besides the gradient descent, there are also the split Bregman method and the iterative convolution-thresholding method to minimize the energy functional. This paper aims at a comprehensive review and summary of the classical and improved models in image segmentation. We detail a general statistical formulation for the variational method and introduce the basic principle, representative works and improvement of several medical image segmentation methods based on variational methods. Subsequently, we show the results and comparisons of some variational methods. We also found that the idea of variational methods can be combined with more recent comparative deep learning to achieve better performance. Finally, the shortcomings and future research directions of medical image segmentation are discussed.
Citation
X. Shu, Z. Li, X. Chang, and D. Yuan, “Variational methods with application to medical image segmentation: A survey,” Neurocomputing, vol. 639, p. 130260, Jul. 2025, doi: 10.1016/J.NEUCOM.2025.130260.
Source
Neurocomputing
Conference
Keywords
Medical image segmentation, Literature survey, Variational methods, Minimize energy functional
Subjects
Source
Publisher
Elsevier