Efficient active contour model for medical image segmentation and correction based on edge and region information. (15th May 2022)

• Record Type: Journal Article
• Authors: Yang, Yunyun; Hou, Xiaoyan; Ren, Huilin
• Is Part Of: Expert systems with applications. Volume 194(2022)
• Publication Date: 2022-05-15
• Subjects: Image segmentation -- Intensity inhomogeneity -- Split Bregman method -- Bias correction -- Medical images -- Active contours; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2021.116436
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: The inhomogeneity of images is always a challenge in the field of image segmentation. Aiming at the problem of segmentation and correction of inhomogeneous images, this paper proposes an improved active contour model based on previous models. Our model inherits the advantage of that, which is robust to images with irregular intensity distribution. Meanwhile, our model can segment and correct images with inhomogeneous intensity at the same time, performing better when segmenting images with inhomogeneous intensity. Moreover, it saves a lot of time because our model has fewer parameters to be tuned. We extend our two-phase model to the multi-phase model and the vector-valued model and prove the convergence of our algorithm. Through a lot of qualitative analysis and quantitative comparison, we find that our model can accurately segment medical images and get homogeneous corrected images. In addition, compared to other models, our model has higher accuracy with fewer iterations. Highlights: Our model can simultaneously segment and correct inhomogeneous images. Less coefficients to be tuned in the framework can greatly save time. Model accurately processes multi-objective magnetic resonance images. The algorithm has been proved to be convergent in theory and practice.