Semi-supervised Left Atrium Segmentation with Mutual Consistency Training

Wu, Yi‐Cheng; Xu, Minfeng; Ge, Zongyuan; Cai, Jianfei; Zhang, Lei

doi:10.1007/978-3-030-87196-3_28

Cited by 98 publications

(59 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The LA dataset consists of 100 gadolinium-enhanced MRI scans, with a fixed split ‡ of 80 samples for training and 20 samples for validation. We report the performance on the validation set for fair comparisons as [24,7,9,19]. On the ACDC dataset, the data split § is also fixed with 70, 10, and 20 patients' scans for training, validation, and testing, respectively.…”

Section: Datasetmentioning

confidence: 99%

“…Following the public methods [24,7,9,19,10] on both datasets, inputs were normalized as zero mean and unit variance. We used the rotation and flip operations to augment data and trained our model via a SGD optimizer with the initial learning rate 0.01 decayed by 10% every 2.5K iterations.…”

Section: Implementation Detailsmentioning

confidence: 99%

“…Its success implies that the local distributional smoothness (LDS) is crucial to leverage abundant unlabeled data for the model training. The low-entropy assumption further constrains that the decision boundary should lie in the lowdensity regions [6,16,19,20]. In other words, the semi-supervised models are expected to output highly confident predictions even without the supervision of corresponding labels, aiming at the inter-class separation.…”

Section: Introductionmentioning

confidence: 99%

“…1, there are many ambiguous pixels near the adhesive edges or low-contrast regions and the amount of labeled data is usually limited. Such concomitant challenges may lead to poor performance of the existing models [24,7,9,10,19]. We further use the synthetic two-moon dataset as a toy example to illustrate this scenario.…”

Section: Introductionmentioning

confidence: 99%

“…Comparisons with five recent methods on the ACDC dataset. All results were re-produced as[24,7,9,10,19] in the identical experimental setting for fair comparisons.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Exploring Smoothness and Class-Separation for Semi-supervised Medical Image Segmentation

Wu¹,

Wu²,

Wu³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Semi-supervised segmentation remains challenging in medical imaging since the amount of annotated medical data is often limited and there are many blurred pixels near the adhesive edges or lowcontrast regions. To address the issues, we advocate to firstly constrain the consistency of samples with and without strong perturbations to apply sufficient smoothness regularization and further encourage the class-level separation to exploit the unlabeled ambiguous pixels for the model training. Particularly, in this paper, we propose the SS-Net for semi-supervised medical image segmentation tasks, via exploring the pixel-level Smoothness and inter-class Separation at the same time. The pixel-level smoothness forces the model to generate invariant results under adversarial perturbations. Meanwhile, the inter-class separation constrains individual class features should approach their corresponding high-quality prototypes, in order to make each class distribution compact and separate different classes. We evaluated our SS-Net against five recent methods on the public LA and ACDC datasets. The experimental results under two semi-supervised settings demonstrate the superiority of our proposed SS-Net, achieving new state-of-the-art (SOTA) performance on both datasets. The codes will be released.

show abstract

Section: Datasetmentioning

confidence: 99%

Section: Implementation Detailsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…Comparisons with five recent methods on the ACDC dataset. All results were re-produced as[24,7,9,10,19] in the identical experimental setting for fair comparisons.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Exploring Smoothness and Class-Separation for Semi-supervised Medical Image Segmentation

Wu¹,

Wu²,

Wu³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

MFA‐ICPS: Semi‐supervised medical image segmentation with improved cross pseudo supervision and multi‐dimensional feature attention

Xia,

He,

Liao

2023

Medical Physics

View full text Add to dashboard Cite

BackgroundIn the medical field, medical image segmentation plays a pivotal role in facilitating disease evaluation and supporting treatment decision‐making for doctors. Recently, deep learning methods have been employed in the field of medical image segmentation. However, the manual annotation of a large number of reliable labels is a costly and time‐consuming process.PurposeTo address this challenge, a semi‐supervised learning framework is required to alleviate the burden of reliable labeling and enhance segmentation accuracy in challenging areas of medical images.MethodsTherefore, this paper presents MFA‐ICPS framework, a semi‐supervised learning framework based on the improved cross pseudo supervision (ICPS) and multi‐dimensional feature attention (MFA) modules. Medical images inevitably contain some noise that may affect the segmentation accuracy, so the proposed framework addresses this challenge by introducing noise disturbance, combining ICPS and MFA modules, and using pseudo‐segmentation maps and MFA maps to maintain the consistency at both the output and feature levels.ResultsIn the experiments, MFA‐ICPS framework accurately obtains the following performances on the left atrial dataset: Dice, Jaccard, 95HD, and ASD values are 90.89%, 83.40%, 6.00 and 1.94 mm, respectively. And on the pancreas‐CT dataset, the following performances are accurately obtained: Dice, Jaccard, 95HD, and ASD values are 79.55%, 66.87%, 7.67 and 1.65 mm, respectively.ConclusionsThe segmentation performance of MFA‐ICPS framework on different medical datasets demonstrates its remarkable capability to significantly enhance medical image segmentation.

show abstract

Multi‐scale consistent self‐training network for semi‐supervised orbital tumor segmentation

Wang,

Jin,

Cheng

et al. 2024

Medical Physics

View full text Add to dashboard Cite

PurposeSegmentation of orbital tumors in CT images is of great significance for orbital tumor diagnosis, which is one of the most prevalent diseases of the eye. However, the large variety of tumor sizes and shapes makes the segmentation task very challenging, especially when the available annotation data is limited.MethodsTo this end, in this paper, we propose a multi‐scale consistent self‐training network (MSCINet) for semi‐supervised orbital tumor segmentation. Specifically, we exploit the semantic‐invariance features by enforcing the consistency between the predictions of different scales of the same image to make the model more robust to size variation. Moreover, we incorporate a new self‐training strategy, which adopts iterative training with an uncertainty filtering mechanism to filter the pseudo‐labels generated by the model, to eliminate the accumulation of pseudo‐label error predictions and increase the generalization of the model.ResultsFor evaluation, we have built two datasets, the orbital tumor binary segmentation dataset (Orbtum‐B) and the orbital multi‐organ segmentation dataset (Orbtum‐M). Experimental results on these two datasets show that our proposed method can both achieve state‐of‐the‐art performance. In our datasets, there are a total of 55 patients containing 602 2D images.ConclusionIn this paper, we develop a new semi‐supervised segmentation method for orbital tumors, which is designed for the characteristics of orbital tumors and exhibits excellent performance compared to previous semi‐supervised algorithms.

show abstract

Semi-supervised Left Atrium Segmentation with Mutual Consistency Training

Cited by 98 publications

References 10 publications

Exploring Smoothness and Class-Separation for Semi-supervised Medical Image Segmentation

Exploring Smoothness and Class-Separation for Semi-supervised Medical Image Segmentation

MFA‐ICPS: Semi‐supervised medical image segmentation with improved cross pseudo supervision and multi‐dimensional feature attention

Multi‐scale consistent self‐training network for semi‐supervised orbital tumor segmentation

Contact Info

Product

Resources

About