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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Deep learning methods typically depend on the availability of labeled data,which is expensive and time-consuming to obtain. Active learning addresses sucheffort by prioritizing which samples are best to annotate in order to maximizethe performance of the task model. While frameworks for active learning havebeen widely explored in the context of classification of natural images, theyhave been only sparsely used in medical image segmentation. The challengeresides in obtaining an uncertainty measure that reveals the best candidatedata for annotation. This paper proposes Test-time Augmentation for ActiveLearning (TAAL), a novel semi-supervised active learning approach forsegmentation that exploits the uncertainty information offered by datatransformations. Our method applies cross-augmentation consistency duringtraining and inference to both improve model learning in a semi-supervisedfashion and identify the most relevant unlabeled samples to annotate next. Inaddition, our consistency loss uses a modified version of the JSD to furtherimprove model performance. By relying on data transformations rather than onexternal modules or simple heuristics typically used in uncertainty-basedstrategies, TAAL emerges as a simple, yet powerful task-agnosticsemi-supervised active learning approach applicable to the medical domain. Ourresults on a publicly-available dataset of cardiac images show that TAALoutperforms existing baseline methods in both fully-supervised andsemi-supervised settings. Our implementation is publicly available onhttps://github.com/melinphd/TAAL.

TAAL: Test-time Augmentation for Active Learning in Medical Image Segmentation