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

Deep learning models usually suffer from domain shift issues, where modelstrained on one source domain do not generalize well to other unseen domains. Inthis work, we investigate the single-source domain generalization problem:training a deep network that is robust to unseen domains, under the conditionthat training data is only available from one source domain, which is common inmedical imaging applications. We tackle this problem in the context ofcross-domain medical image segmentation. Under this scenario, domain shifts aremainly caused by different acquisition processes. We propose a simplecausality-inspired data augmentation approach to expose a segmentation model tosynthesized domain-shifted training examples. Specifically, 1) to make the deepmodel robust to discrepancies in image intensities and textures, we employ afamily of randomly-weighted shallow networks. They augment training imagesusing diverse appearance transformations. 2) Further we show that spuriouscorrelations among objects in an image are detrimental to domain robustness.These correlations might be taken by the network as domain-specific clues formaking predictions, and they may break on unseen domains. We remove thesespurious correlations via causal intervention. This is achieved by resamplingthe appearances of potentially correlated objects independently. The proposedapproach is validated on three cross-domain segmentation tasks: cross-modality(CT-MRI) abdominal image segmentation, cross-sequence (bSSFP-LGE) cardiac MRIsegmentation, and cross-center prostate MRI segmentation. The proposed approachyields consistent performance gains compared with competitive methods whentested on unseen domains.

Causality-inspired Single-source Domain Generalization for Medical Image Segmentation