FedDG: Federated Domain Generalization on Medical Image Segmentation via Episodic Learning in Continuous Frequency Space

@article{Liu2021FedDGFD,
  title={FedDG: Federated Domain Generalization on Medical Image Segmentation via Episodic Learning in Continuous Frequency Space},
  author={Quande Liu and Cheng Chen and Jing Qin and Qi Dou and Pheng-Ann Heng},
  journal={2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2021},
  pages={1013-1023}
}
  • Quande Liu, Cheng Chen, P. Heng
  • Published 10 March 2021
  • Computer Science
  • 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Federated learning allows distributed medical institutions to collaboratively learn a shared prediction model with privacy protection. While at clinical deployment, the models trained in federated learning can still suffer from performance drop when applied to completely unseen hospitals outside the federation. In this paper, we point out and solve a novel problem setting of federated domain generalization (FedDG), which aims to learn a federated model from multiple distributed source domains… 
FedMix: Mixed Supervised Federated Learning for Medical Image Segmentation
TLDR
This work proposes a label-agnostic unified federated learning framework, named FedMix, for medical image segmentation based on mixed image labels, which outperforms the state-of-the-art methods by a large margin.
Federated Semi-supervised Medical Image Classification via Inter-client Relation Matching
TLDR
This paper studies a practical yet challenging FL problem, named Federated Semi-supervised Learning (FSSL), which aims to learn a federated model by jointly utilizing the data from both labeled and unlabeled clients, and presents a novel approach for this problem, which improves over traditional consistency regularization mechanism with a new inter-client relation matching scheme.
Federated Cross Learning for Medical Image Segmentation
TLDR
The experimental results show that the proposed FedCross training strategy outperforms the mainstream FL methods on non-iid data, and can further provide a quantitative estimation of the model uncertainty, demonstrating the ef-fectiveness and clinical significance of the designs.
Federated Learning with Domain Generalization
TLDR
FedADG is proposed to equip federated learning with domain generalization capability and has better performance than most of the previous solutions even if they have an additional advantage that allows centralized data access.
Federated Multi-organ Segmentation with Partially Labeled Data
TLDR
This work uses a novel federated multi-encoding U-Net (Fed-MENU) method to effectively obtain a federated learning model using the partially labeled datasets with superior performance to other models trained by either localized or centralized learning methods.
IOP-FL: Inside-Outside Personalization for Federated Medical Image Segmentation
TLDR
An extensive experimental results on two medical image segmentation tasks present improvements over SOTA methods on both inside and outside personalization, demonstrating the great potential of the IOP-FL scheme for clinical practice.
Closing the Generalization Gap of Cross-silo Federated Medical Image Segmentation
TLDR
This work proposes a novel training framework FedSM to avoid the client drift issue and successfully close the generalization gap compared with the centralized training for medical image segmentation tasks for the first time.
FRAug: Tackling Federated Learning with Non-IID Features via Representation Augmentation
TLDR
This work addresses the recently proposed feature shift problem where the clients have different feature distributions while the label distribution is the same, and substantially outperforms the current state-of-the-art FL methods for non-IID features, including PartialFed and FedBN.
Federated Learning and Differential Privacy for Medical Image Analysis
TLDR
This work empirically compares the performance of private, distributed training to conventional training and demonstrates that distributed training can achieve similar performance with strong privacy guarantees and indicates that differentially private federated learning is a viable and reliable framework for the collaborative development of machine learning models in medical image analysis.
...
...

References

SHOWING 1-10 OF 65 REFERENCES
Federated Adversarial Domain Adaptation
TLDR
This work presents a principled approach to the problem of federated domain adaptation, which aims to align the representations learned among the different nodes with the data distribution of the target node.
Synthetic Learning: Learn From Distributed Asynchronized Discriminator GAN Without Sharing Medical Image Data
TLDR
A data privacy-preserving and communication efficient distributed GAN learning framework named Distributed Asynchronized Discriminator GAN (AsynDGAN), which aims to train a central generator learns from distributed discriminator, and use the generated synthetic image solely to train the segmentation model.
Privacy-preserving Federated Brain Tumour Segmentation
TLDR
The feasibility of applying differential-privacy techniques to protect the patient data in a federated learning setup for brain tumour segmentation on the BraTS dataset is investigated and there is a trade-off between model performance and privacy protection costs.
Generalizing Deep Learning for Medical Image Segmentation to Unseen Domains via Deep Stacked Transformation
TLDR
A deep stacked transformation approach for domain generalization that can be generalized to the design of highly robust deep segmentation models for clinical deployment and reaches the performance of state-of theart fully supervised models that are trained and tested on their source domains.
Federated Visual Classification with Real-World Data Distribution
TLDR
Two new large-scale datasets for species and landmark classification are introduced, with realistic per-user data splits that simulate real-world edge learning scenarios, and two new algorithms are developed that intelligently resample and reweight over the client pool, bringing large improvements in accuracy and stability in training.
Federated Learning in Distributed Medical Databases: Meta-Analysis of Large-Scale Subcortical Brain Data
TLDR
A federated learning framework for securely accessing and meta-analyzing any biomedical data without sharing individual information is proposed and applied to multi-centric, multi-database studies including ADNI, PPMI, MIRIAD and UK Biobank, showing the potential of the approach for further applications in distributed analysis ofMulti-centric cohorts.
Multi-Institutional Deep Learning Modeling Without Sharing Patient Data: A Feasibility Study on Brain Tumor Segmentation
TLDR
This study introduces the first use of federated learning for multi-institutional collaboration, enabling deep learning modeling without sharing patient data, and demonstrates that the performance of Federated semantic segmentation models on multimodal brain scans is similar to that of models trained by sharing data.
Federated Simulation for Medical Imaging
TLDR
This work introduces a physics-driven generative approach that consists of two learnable neural modules: a module that synthesizes 3D cardiac shapes along with their materials, and a CT simulator that renders these into realistic 3D CT Volumes, with annotations.
Federated Learning for Breast Density Classification: A Real-World Implementation
TLDR
This study investigates the use of federated learning (FL) to build medical imaging classification models in a real-world collaborative setting and shows that despite substantial differences among the datasets from all sites and without centralizing data, it can successfully train AI models in federation.
...
...