Multi-agent medical image segmentation: A survey

Bennai, Mohamed Tahar; Guessoum, Zahia; Mazouzi, Smaine; Cormier, Stéphane; Mezghiche, Mohamed

doi:10.1016/j.cmpb.2023.107444

Cited by 10 publications

(3 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, future assisted diagnosis and treatment systems will be dominated by multi-sample-based unsupervised data enhancement techniques, which are more dedicated to the reduction of parameter space and the expansion of the number of potential strategies and eventually transformed into a combination of unsupervised methods based on the generation of data and the learning of enhancement operations [20][21]. With the continuous innovation of image processing techniques, especially deep learning techniques, many areas with limited development in the past or emerging development areas are developing rapidly [22][23]. Considering how to break through the limitations of inherent concepts and combining emerging technologies with traditional concepts will create more new possibilities and solve some difficult problems that cannot be solved by traditional methods.…”

Section: Introductionmentioning

confidence: 99%

Research on Medical Image Enhancement Method Based on Conditional Entropy Generative Adversarial Networks

2024

Applied Mathematics and Nonlinear Sciences

View full text Add to dashboard Cite

This study proposes an image enhancement method combining conditional entropy and generative adversarial network, aiming to improve the image quality while avoiding overfitting through the negative training of dependent generative adversarial network and introducing dependent entropy distance loss. Through NIQMC, NIQE and BTMQI evaluation indexes, this paper evaluates the effects of different parameter combinations and image chunk sizes on the enhancement results. It utilizes information entropy as an evaluation index to measure the impact of conditional entropy distance loss. The effectiveness of adversarial learning and conditional entropy in image enhancement is verified by comparing the experimental results. The experiments show that the system can achieve the best image quality of SSIM=0.9852, PSNR=27.58, and SNROI=21.34 with the parameters S=50 and R=4.0%, indicating that the method can effectively retain the detailed information and realism of the Image while enhancing the clarity of the Image, demonstrating a significant performance advantage.

show abstract

Section: Introductionmentioning

confidence: 99%

Research on Medical Image Enhancement Method Based on Conditional Entropy Generative Adversarial Networks

2024

Applied Mathematics and Nonlinear Sciences

View full text Add to dashboard Cite

show abstract

“…Image segmentation is a challenging and complicated task in the field of image processing and has a wide range of applications in computer vision, such as medical image analysis, autonomous driving, remote sensing, security and protection monitoring [1][2][3]. The major task of image segmentation is to divide a prescribed image into several nonoverlapping and disjoint regions according to characteristics such as color, gray level (intensity) and geometric shape.…”

Section: Introductionmentioning

confidence: 99%

An Efficient Local Radial Basis Function Method for Image Segmentation Based on the Chan–Vese Model

Qiu,

Lin,

Zhao

2024

CMES

View full text Add to dashboard Cite

In this paper, we consider the Chan-Vese (C-V) model for image segmentation and obtain its numerical solution accurately and efficiently. For this purpose, we present a local radial basis function method based on a Gaussian kernel (GA-LRBF) for spatial discretization. Compared to the standard radial basis function method, this approach consumes less CPU time and maintains good stability because it uses only a small subset of points in the whole computational domain. Additionally, since the Gaussian function has the property of dimensional separation, the GA-LRBF method is suitable for dealing with isotropic images. Finally, a numerical scheme that couples GA-LRBF with the fourth-order Runge-Kutta method is applied to the C-V model, and a comparison of some numerical results demonstrates that this scheme achieves much more reliable image segmentation.

show abstract

“…With society's burgeoning reliance on image information, image segmentation has emerged as a pivotal task in the realm of computer vision and is now extensively utilized across myriad sectors [1]. In realms such as medical imaging, autonomous driving, and industrial inspection, the significance of image segmentation technology in information extraction and analysis has been well-established [2][3][4][5][6][7]. However, conventional image segmentation approaches, in the face of escalating data volumes and dimensionality, have shown shortcomings in efficiency and accuracy, especially in high-dimensional data flow environments [8,9].…”

Section: Introductionmentioning

confidence: 99%

Optimization of Deep Learning Algorithms for Image Segmentation in High-Dimensional Data Environments

2023

View full text Add to dashboard Cite

As image segmentation tasks become increasingly intricate within high-dimensional data flow environments, conventional segmentation techniques are challenged in delivering both efficiency and precision. In this context, the problem of image segmentation under highdimensional data flux was examined. Depth skip connections, inspired by the U-Net architecture, were introduced, harnessing the superior feature extraction capabilities of deep encoders and enabling the formulation of a lightweight model structure. Furthermore, an equilibrium between Binary Cross-Entropy (BCE) loss and Dice loss was established, targeting enhanced accuracy in small object segmentation tasks within such data-intensive settings. These innovations not only augment algorithmic accuracy and resilience but also provide pivotal contributions to ongoing research in the image segmentation realm. The methodologies delineated herein present a refined approach to image segmentation, revealing significant potential for application in pivotal sectors, including medical image analysis and autonomous vehicular navigation.

show abstract

Multi-agent medical image segmentation: A survey

Cited by 10 publications

References 37 publications

Research on Medical Image Enhancement Method Based on Conditional Entropy Generative Adversarial Networks

Research on Medical Image Enhancement Method Based on Conditional Entropy Generative Adversarial Networks

An Efficient Local Radial Basis Function Method for Image Segmentation Based on the Chan–Vese Model

Optimization of Deep Learning Algorithms for Image Segmentation in High-Dimensional Data Environments

Contact Info

Product

Resources

About