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

Annotation scarcity has become a major obstacle for training powerfuldeep-learning models for medical image segmentation, restricting theirdeployment in clinical scenarios. To address it, semi-supervised learning byexploiting abundant unlabeled data is highly desirable to boost the modeltraining. However, most existing works still focus on limited medical tasks andunderestimate the potential of learning across diverse tasks and multipledatasets. Therefore, in this paper, we introduce a \textbf{Ver}satile\textbf{Semi}-supervised framework (VerSemi) to point out a new perspectivethat integrates various tasks into a unified model with a broad label space, toexploit more unlabeled data for semi-supervised medical image segmentation.Specifically, we introduce a dynamic task-prompted design to segment varioustargets from different datasets. Next, this unified model is used to identifythe foreground regions from all labeled data, to capture cross-datasetsemantics. Particularly, we create a synthetic task with a cutmix strategy toaugment foreground targets within the expanded label space. To effectivelyutilize unlabeled data, we introduce a consistency constraint. This involvesaligning aggregated predictions from various tasks with those from thesynthetic task, further guiding the model in accurately segmenting foregroundregions during training. We evaluated our VerSemi model on four publicbenchmarking datasets. Extensive experiments demonstrated that VerSemi canconsistently outperform the second-best method by a large margin (e.g., anaverage 2.69\% Dice gain on four datasets), setting new SOTA performance forsemi-supervised medical image segmentation. The code will be released.

Segment Together: A Versatile Paradigm for Semi-Supervised Medical Image Segmentation