Ca-Mt: A Self-Ensembling Model for Semi-Supervised Cardiac Segmentation with Elliptical Descriptor Based Contour-Aware

Xu, An; Wang, Shaoyu; Ye, Shaoping; Fan, Jie; Shi, Xiaohu; Xia, Xiaoling

doi:10.1109/isbi52829.2022.9761666

Cited by 4 publications

(4 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instead of teacher and student models sharing the training weights, in the MT framework, the teacher's weights are updated using the exponential moving average (EMA) of the student model. 208 Following its recent success in computer vision, the MT framework has been successfully employed in various medical segmentation algorithms for brain 150,153,154 and cardiac 157,160,162 MRI segmentation.…”

Section: Student-teacher Methodologymentioning

confidence: 99%

A review of self‐supervised, generative, and few‐shot deep learning methods for data‐limited magnetic resonance imaging segmentation

Liu,

Kainth,

Zhou

et al. 2024

NMR in Biomedicine

View full text Add to dashboard Cite

Magnetic resonance imaging (MRI) is a ubiquitous medical imaging technology with applications in disease diagnostics, intervention, and treatment planning. Accurate MRI segmentation is critical for diagnosing abnormalities, monitoring diseases, and deciding on a course of treatment. With the advent of advanced deep learning frameworks, fully automated and accurate MRI segmentation is advancing. Traditional supervised deep learning techniques have advanced tremendously, reaching clinical‐level accuracy in the field of segmentation. However, these algorithms still require a large amount of annotated data, which is oftentimes unavailable or impractical. One way to circumvent this issue is to utilize algorithms that exploit a limited amount of labeled data. This paper aims to review such state‐of‐the‐art algorithms that use a limited number of annotated samples. We explain the fundamental principles of self‐supervised learning, generative models, few‐shot learning, and semi‐supervised learning and summarize their applications in cardiac, abdomen, and brain MRI segmentation. Throughout this review, we highlight algorithms that can be employed based on the quantity of annotated data available. We also present a comprehensive list of notable publicly available MRI segmentation datasets. To conclude, we discuss possible future directions of the field—including emerging algorithms, such as contrastive language‐image pretraining, and potential combinations across the methods discussed—that can further increase the efficacy of image segmentation with limited labels.

show abstract

Section: Student-teacher Methodologymentioning

confidence: 99%

A review of self‐supervised, generative, and few‐shot deep learning methods for data‐limited magnetic resonance imaging segmentation

Liu,

Kainth,

Zhou

et al. 2024

NMR in Biomedicine

View full text Add to dashboard Cite

show abstract

“…The University Hospital of Dijon (France) provides 150 short‐axis 2D cine‐MRI imaging samples. We extract data from 100 samples out of the whole dataset and split them into 70 samples for the training set, 10 samples for the validation set, and 20 samples for the testing set 31 . Unlike the previous 3D image segmentation tasks, ACDC is a 2D dataset where we have modified the baseline V‐Net to a 2D U‐Net model.…”

Section: Methodsmentioning

confidence: 99%

“…We extract data from 100 samples out of the whole dataset and split them into 70 samples for the training set, 10 samples for the validation set, and 20 samples for the testing set. 31 Unlike the previous 3D image segmentation tasks, ACDC is a 2D dataset where we have modified the baseline V-Net to a 2D U-Net model. Its object is to segment the myocardium, left ventricle, and right ventricle from 2D MR slices separately.…”

Section: Databasementioning

confidence: 99%

Multidimensional perturbed consistency learning for semi‐supervised medical image segmentation

Yuan,

Zhao,

Qin

et al. 2024

Int J Imaging Syst Tech

View full text Add to dashboard Cite

In this article, we propose a novel multidimensional perturbed consistency network (MPCNet) for more accurate semi‐supervised medical image segmentation. Specifically, we develop a multidimensional perturbation by considering the noise itself, the target object and the overall spatial architecture. This type of perturbation can disrupt the propagation process of abstract features over a wide range, enabling the model to learn the distribution of comprehensive data. In addition, we design a shared encoder to extract multi‐scale features. After subjecting these features to multidimensional perturbation, a consistency constraint is applied between different results output by three independent decoders. This constraint aims to minimize the statistical differences between these results and effectively leverage unlabeled data. Experimental results on the public LA, Pancreas‐CT and ACDC datasets demonstrate that our method outperforms recent SOTA semi‐supervised learning methods in terms of various metrics. Our code is released publicly at https://github.com/yuanenze123/MPC-Net.

show abstract

“…Except encouraging consistency on network segmentation results directly, generative consistency [59] is proposed through a generation network that reconstructs medical images from its predictions of the segmentation network. Xu et al [60] propose contour consistency and utilize Fourier series which contained a series of harmonics as an elliptical descriptor. Through minimizing the L2 distance of the parameters between the student and the teacher branch, the model is equipped with shape awareness.…”

Section: Unsupervised Regularization With Consistency Learningmentioning

confidence: 99%

Learning with Limited Annotations: A Survey on Deep Semi-Supervised Learning for Medical Image Segmentation

Jiao¹,

Zhang²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Medical image segmentation is a fundamental and critical step in many image-guided clinical approaches. Recent success of deep learning-based segmentation methods usually relies on a large amount of labeled data, which is particularly difficult and costly to obtain especially in the medical imaging domain where only experts can provide reliable and accurate annotations. Semi-supervised learning has emerged as an appealing strategy and been widely applied to medical image segmentation tasks to train deep models with limited annotations. In this paper, we present a comprehensive review of recently proposed semi-supervised learning methods for medical image segmentation and summarized both the technical novelties and empirical results. Furthermore, we analyze and discuss the limitations and several unsolved problems of existing approaches. We hope this review could inspire the research community to explore solutions for this challenge and further promote the developments in medical image segmentation field.

show abstract

Ca-Mt: A Self-Ensembling Model for Semi-Supervised Cardiac Segmentation with Elliptical Descriptor Based Contour-Aware

Cited by 4 publications

References 11 publications

A review of self‐supervised, generative, and few‐shot deep learning methods for data‐limited magnetic resonance imaging segmentation

A review of self‐supervised, generative, and few‐shot deep learning methods for data‐limited magnetic resonance imaging segmentation

Multidimensional perturbed consistency learning for semi‐supervised medical image segmentation

Learning with Limited Annotations: A Survey on Deep Semi-Supervised Learning for Medical Image Segmentation

Contact Info

Product

Resources

About