FedDM: Federated Weakly Supervised Segmentation via Annotation Calibration and Gradient De-Conflicting

Zhu, Meilu; Chen, Zhen; Yuan, Yixuan

doi:10.1109/tmi.2023.3235757

Cited by 10 publications

(9 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For a comprehensive evaluation of FedA 3 I on addressing heterogeneous annotation noise, a set of SOTA methods are selected for comparison, including GCE (Zhang and Sabuncu 2018), SCE (Wang et al 2019), ELR (Liu et al 2020), ADELE (Liu et al 2022), RMD (Fang et al 2023), NR-Dice (Wang et al 2020), FedProx , FedMix (Wicaksana et al 2022), FedDM (Zhu, Chen, and Yuan 2023), FedCorr (Xu et al 2022b), and FedNoRo (Wu et al 2023a). The introduction and implementation details of these methods can be found in Appendix B.…”

Section: Comparison With Sota Methodsmentioning

confidence: 99%

“…It confirms the necessity of specific designs for the two-level Non-IID annotation noise in FMIS. In addition, different from those methods introducing complex correction (Xu et al 2022b;Liu et al 2022;Zhu, Chen, and Yuan 2023) or regularization (Liu et al 2020;Wu et al 2023a), no additional training overhead is introduced in FedA 3 I, making it more easy-todeploy in practice.…”

Section: Comparison With Sota Methodsmentioning

confidence: 99%

“…For simplification, this paper focuses on the typical binary segmentation problem 1 following previous studies (Wicaksana et al 2022;Zhu, Chen, and Yuan 2023). Given K participants in FL, each client k holds a private dataset…”

Section: Preliminariesmentioning

confidence: 99%

“…Despite extensive studies on addressing related issues like 1. Noisy labels for classification (Xu et al 2022b;Wu et al 2023a) and weak annotations for segmentation (Wicaksana et al 2022;Zhu, Chen, and Yuan 2023) in FL, 2. Noisy annotations for segmentation in centralized learning (CL) (Liu et al 2022;Yao et al 2023), noisy annotations in FMIS may not be well addressed based on existing methods.…”

Section: Introductionmentioning

confidence: 99%

“…In this paper, we, for the first time, identify and formulate this new FMIS problem, i.e., how to train a federated segmentation model with noisy pixel-wise annotations. Different from weak annotations like bounding boxes studied in previous FMIS works (Wicaksana et al 2022;Zhu, Chen, and Yuan 2023), noisy annotations studied in this paper are more common and realistic in clinical practice and thus more worthy of being addressed. To formulate annotation noise, inspired by the real-world annotating process of delineating ROI contours, we first propose a generic model to generate noisy annotations named contour evolution model (CEM).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

FedA3I: Annotation Quality-Aware Aggregation for Federated Medical Image Segmentation against Heterogeneous Annotation Noise

Wu,

Sun,

Yan

et al. 2024

AAAI

View full text Add to dashboard Cite

Federated learning (FL) has emerged as a promising paradigm for training segmentation models on decentralized medical data, owing to its privacy-preserving property. However, existing research overlooks the prevalent annotation noise encountered in real-world medical datasets, which limits the performance ceilings of FL. In this paper, we, for the first time, identify and tackle this problem. For problem formulation, we propose a contour evolution for modeling non-independent and identically distributed (Non-IID) noise across pixels within each client and then extend it to the case of multi-source data to form a heterogeneous noise model (i.e., Non-IID annotation noise across clients). For robust learning from annotations with such two-level Non-IID noise, we emphasize the importance of data quality in model aggregation, allowing high-quality clients to have a greater impact on FL. To achieve this, we propose Federated learning with Annotation quAlity-aware AggregatIon, named FedA3I, by introducing a quality factor based on client-wise noise estimation. Specifically, noise estimation at each client is accomplished through the Gaussian mixture model and then incorporated into model aggregation in a layer-wise manner to up-weight high-quality clients. Extensive experiments on two real-world medical image segmentation datasets demonstrate the superior performance of FedA3I against the state-of-the-art approaches in dealing with cross-client annotation noise. The code is available at https://github.com/wnn2000/FedAAAI.

show abstract

Section: Comparison With Sota Methodsmentioning

confidence: 99%

Section: Comparison With Sota Methodsmentioning

confidence: 99%

Section: Preliminariesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

FedA3I: Annotation Quality-Aware Aggregation for Federated Medical Image Segmentation against Heterogeneous Annotation Noise

Wu,

Sun,

Yan

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

Rethinking Polyp Segmentation From An Out-of-distribution Perspective

Ji,

Zhang,

Campbell

et al. 2024

Mach. Intell. Res.

View full text Add to dashboard Cite

Unlike existing fully-supervised approaches, we rethink colorectal polyp segmentation from an out-of-distribution perspective with a simple but effective self-supervised learning approach. We leverage the ability of masked autoencoders–self-supervised vision transformers trained on a reconstruction task–to learn in-distribution representations, here, the distribution of healthy colon images. We then perform out-of-distribution reconstruction and inference, with feature space standardisation to align the latent distribution of the diverse abnormal samples with the statistics of the healthy samples. We generate per-pixel anomaly scores for each image by calculating the difference between the input and reconstructed images and use this signal for out-of-distribution (i.e., polyp) segmentation. Experimental results on six benchmarks show that our model has excellent segmentation performance and generalises across datasets. Our code is publicly available at https://github.com/GewelsJI/Polyp-OOD.

show abstract

Emerging trends in federated learning: from model fusion to federated X learning

Ji,

Tan,

Saravirta

et al. 2024

Int. J. Mach. Learn. & Cyber.

View full text Add to dashboard Cite

Federated learning is a new learning paradigm that decouples data collection and model training via multi-party computation and model aggregation. As a flexible learning setting, federated learning has the potential to integrate with other learning frameworks. We conduct a focused survey of federated learning in conjunction with other learning algorithms. Specifically, we explore various learning algorithms to improve the vanilla federated averaging algorithm and review model fusion methods such as adaptive aggregation, regularization, clustered methods, and Bayesian methods. Following the emerging trends, we also discuss federated learning in the intersection with other learning paradigms, termed federated X learning, where X includes multitask learning, meta-learning, transfer learning, unsupervised learning, and reinforcement learning. In addition to reviewing state-of-the-art studies, this paper also identifies key challenges and applications in this field, while also highlighting promising future directions.

show abstract

FedDM: Federated Weakly Supervised Segmentation via Annotation Calibration and Gradient De-Conflicting

Cited by 10 publications

References 41 publications

FedA3I: Annotation Quality-Aware Aggregation for Federated Medical Image Segmentation against Heterogeneous Annotation Noise

FedA3I: Annotation Quality-Aware Aggregation for Federated Medical Image Segmentation against Heterogeneous Annotation Noise

Rethinking Polyp Segmentation From An Out-of-distribution Perspective

Emerging trends in federated learning: from model fusion to federated X learning

Contact Info

Product

Resources

About