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

Self-training based unsupervised domain adaptation (UDA) has shown greatpotential to address the problem of domain shift, when applying a trained deeplearning model in a source domain to unlabeled target domains. However, whilethe self-training UDA has demonstrated its effectiveness on discriminativetasks, such as classification and segmentation, via the reliable pseudo-labelselection based on the softmax discrete histogram, the self-training UDA forgenerative tasks, such as image synthesis, is not fully investigated. In thiswork, we propose a novel generative self-training (GST) UDA framework withcontinuous value prediction and regression objective for cross-domain imagesynthesis. Specifically, we propose to filter the pseudo-label with anuncertainty mask, and quantify the predictive confidence of generated imageswith practical variational Bayes learning. The fast test-time adaptation isachieved by a round-based alternative optimization scheme. We validated ourframework on the tagged-to-cine magnetic resonance imaging (MRI) synthesisproblem, where datasets in the source and target domains were acquired fromdifferent scanners or centers. Extensive validations were carried out to verifyour framework against popular adversarial training UDA methods. Results showthat our GST, with tagged MRI of test subjects in new target domains, improvedthe synthesis quality by a large margin, compared with the adversarial trainingUDA methods.

Generative Self-training for Cross-domain Unsupervised Tagged-to-Cine MRI Synthesis