English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Federated learning (FL) enables multiple sites to collaboratively trainpowerful deep models without compromising data privacy and security. Thestatistical heterogeneity (e.g., non-IID data and domain shifts) is a primaryobstacle in FL, impairing the generalization performance of the global model.Weakly supervised segmentation, which uses sparsely-grained (i.e., point-,bounding box-, scribble-, block-wise) supervision, is increasingly being paidattention to due to its great potential of reducing annotation costs. However,there may exist label heterogeneity, i.e., different annotation forms acrosssites. In this paper, we propose a novel personalized FL framework for medicalimage segmentation, named FedICRA, which uniformly leverages heterogeneous weaksupervision via adaptIve Contrastive Representation and Aggregation.Concretely, to facilitate personalized modeling and to avoid confusion, achannel selection based site contrastive representation module is employed toadaptively cluster intra-site embeddings and separate inter-site ones. Toeffectively integrate the common knowledge from the global model with theunique knowledge from each local model, an adaptive aggregation module isapplied for updating and initializing local models at the element level.Additionally, a weakly supervised objective function that leverages amultiscale tree energy loss and a gated CRF loss is employed to generate moreprecise pseudo-labels and further boost the segmentation performance. Throughextensive experiments on two distinct medical image segmentation tasks ofdifferent modalities, the proposed FedICRA demonstrates overwhelmingperformance over other state-of-the-art personalized FL methods. Itsperformance even approaches that of fully supervised training on centralizeddata. Our code and data are available at https://github.com/llmir/FedICRA.

Unifying and Personalizing Weakly-supervised Federated Medical Image Segmentation via Adaptive Representation and Aggregation