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

Unsupervised domain adaptation (UDA) has increasingly gained interests forits capacity to transfer the knowledge learned from a labeled source domain toan unlabeled target domain. However, typical UDA methods require concurrentaccess to both the source and target domain data, which largely limits itsapplication in medical scenarios where source data is often unavailable due toprivacy concern. To tackle the source data-absent problem, we present a noveltwo-stage source-free domain adaptation (SFDA) framework for medical imagesegmentation, where only a well-trained source segmentation model and unlabeledtarget data are available during domain adaptation. Specifically, in theprototype-anchored feature alignment stage, we first utilize the weights of thepre-trained pixel-wise classifier as source prototypes, which preserve theinformation of source features. Then, we introduce the bi-directional transportto align the target features with class prototypes by minimizing its expectedcost. On top of that, a contrastive learning stage is further devised toutilize those pixels with unreliable predictions for a more compact targetfeature distribution. Extensive experiments on a cross-modality medicalsegmentation task demonstrate the superiority of our method in large domaindiscrepancy settings compared with the state-of-the-art SFDA approaches andeven some UDA methods. Code is available athttps://github.com/CSCYQJ/MICCAI23-ProtoContra-SFDA.

Source-Free Domain Adaptation for Medical Image Segmentation via Prototype-Anchored Feature Alignment and Contrastive Learning