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 neural networks are commonly used for automated medical imagesegmentation, but models will frequently struggle to generalize well acrossdifferent imaging modalities. This issue is particularly problematic due to thelimited availability of annotated data, making it difficult to deploy thesemodels on a larger scale. To overcome these challenges, we propose a newsemi-supervised training strategy called MoDATTS. Our approach is designed foraccurate cross-modality 3D tumor segmentation on unpaired bi-modal datasets. Animage-to-image translation strategy between imaging modalities is used toproduce annotated pseudo-target volumes and improve generalization to theunannotated target modality. We also use powerful vision transformerarchitectures and introduce an iterative self-training procedure to furtherclose the domain gap between modalities. MoDATTS additionally allows thepossibility to extend the training to unannotated target data by exploitingimage-level labels with an unsupervised objective that encourages the model toperform 3D diseased-to-healthy translation by disentangling tumors from thebackground. The proposed model achieves superior performance compared to othermethods from participating teams in the CrossMoDA 2022 challenge, as evidencedby its reported top Dice score of 0.87+/-0.04 for the VS segmentation. MoDATTSalso yields consistent improvements in Dice scores over baselines on across-modality brain tumor segmentation task composed of four differentcontrasts from the BraTS 2020 challenge dataset, where 95% of a targetsupervised model performance is reached. We report that 99% and 100% of thismaximum performance can be attained if 20% and 50% of the target data isadditionally annotated, which further demonstrates that MoDATTS can beleveraged to reduce the annotation burden.

Image-level supervision and self-training for transformer-based cross-modality tumor segmentation