Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
102 tokens/sec
GPT-4o
59 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
50 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

FedA3I: Annotation Quality-Aware Aggregation for Federated Medical Image Segmentation against Heterogeneous Annotation Noise (2312.12838v2)

Published 20 Dec 2023 in cs.LG and cs.CV

Abstract: 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 FedA$3$I against the state-of-the-art approaches in dealing with cross-client annotation noise. The code is available at https://github.com/wnn2000/FedAAAI.

View on arXiv
Definition Search Book Streamline Icon: https://streamlinehq.com
References (50)
  1. Dataset of breast ultrasound images. Data in brief, 28: 104863.
  2. Understanding and improving early stopping for learning with noisy labels. In NeurIPS, 24392–24403.
  3. Introduction of human-centric AI assistant to aid radiologists for multimodal breast image classification. Int. J. Hum. Comput. Stud., 150: 102607.
  4. Beyond class-conditional assumption: A primary attempt to combat instance-dependent label noise. In AAAI, 11442–11450.
  5. Skin lesion analysis toward melanoma detection: A challenge at the 2017 international symposium on biomedical imaging (isbi), hosted by the international skin imaging collaboration (isic). In ISBI, 168–172. IEEE.
  6. Federated deep learning for detecting COVID-19 lung abnormalities in CT: a privacy-preserving multinational validation study. npj Digit. Medicine, 4(1): 60.
  7. Reliable Mutual Distillation for Medical Image Segmentation under Imperfect Annotations. IEEE Trans. Medical Imaging.
  8. CE-Net: Context encoder network for 2D medical image segmentation. IEEE Trans. Medical Imaging, 38(10): 2281–2292.
  9. Co-teaching: Robust training of deep neural networks with extremely noisy labels. In NeurIPS.
  10. Generalizable heterogeneous federated cross-correlation and instance similarity learning. IEEE Trans. Pattern Anal. Mach. Intell.
  11. Federated Learning for Generalization, Robustness, Fairness: A Survey and Benchmark. arXiv:2311.06750.
  12. Learning calibrated medical image segmentation via multi-rater agreement modeling. In CVPR, 12341–12351.
  13. Fair Federated Medical Image Segmentation via Client Contribution Estimation. In CVPR, 16302–16311.
  14. IOP-FL: Inside-outside personalization for federated medical image segmentation. IEEE Trans. Medical Imaging.
  15. Towards federated learning against noisy labels via local self-regularization. In CIKM, 862–873.
  16. DivideMix: Learning with Noisy Labels as Semi-supervised Learning. In ICLR.
  17. Superpixel-guided iterative learning from noisy labels for medical image segmentation. In MICCAI, 525–535. Springer.
  18. Federated optimization in heterogeneous networks. In MLSys, 429–450.
  19. FedBN: Federated learning on non-iid features via local batch normalization. In ICLR.
  20. RSCFed: Random Sampling Consensus Federated Semi-supervised Learning. In CVPR, 10154–10163.
  21. FedDG: Federated domain generalization on medical image segmentation via episodic learning in continuous frequency space. In CVPR, 1013–1023.
  22. Adaptive early-learning correction for segmentation from noisy annotations. In CVPR, 2606–2616.
  23. Early-learning regularization prevents memorization of noisy labels. In NeurIPS, volume 33, 20331–20342.
  24. Layer-wised model aggregation for personalized federated learning. In CVPR, 10092–10101.
  25. Communication-efficient learning of deep networks from decentralized data. In AISTATS, 1273–1282.
  26. V-net: Fully convolutional neural networks for volumetric medical image segmentation. In 3DV, 565–571. Ieee.
  27. The future of digital health with federated learning. npj Digit. Medicine, 3(1): 119.
  28. U-Net: Convolutional networks for biomedical image segmentation. In MICCAI, 234–241. Springer.
  29. Tversky loss function for image segmentation using 3D fully convolutional deep networks. In Machine Learning in Medical Imaging - 8th International Workshop, 379–387. Springer.
  30. UNeXt: MLP-based rapid medical image segmentation network. In MICCAI, 23–33. Springer.
  31. A noise-robust framework for automatic segmentation of COVID-19 pneumonia lesions from CT images. IEEE Trans. Medical Imaging, 39(8): 2653–2663.
  32. FedDP: Dual personalization in federated medical image segmentation. IEEE Trans. Medical Imaging.
  33. Personalizing federated medical image segmentation via local calibration. In ECCV, 456–472.
  34. Symmetric cross entropy for robust learning with noisy labels. In CVPR, 322–330.
  35. FedMix: Mixed Supervised Federated Learning for Medical Image Segmentation. IEEE Trans. Medical Imaging.
  36. FedNoRo: Towards Noise-Robust Federated Learning by Addressing Class Imbalance and Label Noise Heterogeneity. In IJCAI.
  37. FedIIC: Towards Robust Federated Learning for Class-Imbalanced Medical Image Classification. In MICCAI.
  38. Closing the generalization gap of cross-silo federated medical image segmentation. In CVPR, 20866–20875.
  39. FedCorr: Multi-Stage Federated Learning for Label Noise Correction. In CVPR, 10184–10193.
  40. Federated Multi-organ Segmentation with Inconsistent Labels. IEEE Trans. Medical Imaging.
  41. Variation-aware federated learning with multi-source decentralized medical image data. IEEE J. Biomed. Health Informatics, 25(7): 2615–2628.
  42. Learning to Segment from Noisy Annotations: A Spatial Correction Approach. In ICLR.
  43. Automated breast ultrasound lesions detection using convolutional neural networks. IEEE J. Biomed. Health Informatics, 22(4): 1218–1226.
  44. How does disagreement help generalization against label corruption? In ICML, 7164–7173. PMLR.
  45. Disentangling human error from ground truth in segmentation of medical images. In NeurIPS, volume 33, 15750–15762.
  46. Robust medical image segmentation from non-expert annotations with tri-network. In MICCAI, 249–258. Springer.
  47. BUSIS: a benchmark for breast ultrasound image segmentation. In Healthcare, volume 10, 729. MDPI.
  48. Generalized cross entropy loss for training deep neural networks with noisy labels. In NeurIPS.
  49. Pick-and-learn: Automatic quality evaluation for noisy-labeled image segmentation. In MICCAI, 576–584. Springer.
  50. FedDM: Federated Weakly Supervised Segmentation via Annotation Calibration and Gradient De-conflicting. IEEE Trans. Medical Imaging.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (4)
  1. Nannan Wu (12 papers)
  2. Zhaobin Sun (3 papers)
  3. Zengqiang Yan (20 papers)
  4. Li Yu (193 papers)
Citations (3)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now