What question did this study set out to answer?

To establish an annotation-efficient semi-supervised segmentation framework by integrating structured medical text priors with a dual-branch vision model.

July 1, 2026Meta-RadiologyOpen Access

Voxel-Level Text-Visual Alignment with Discrepancy-Aware Fusion for Semi-Supervised Multi-Organ Segmentation

View Full Paper

Ask AI

Bookmark

Authors

WCWenghao Chen SLSheng Lian JLJiayao Liu

Discussion

Member takes

Overview

Computational study demonstrates improved semi-supervised segmentation accuracy across multiple organs in CT and MRI scans, highlighting potential for annotation-efficient clinical workflows.

Key Points

To establish an annotation-efficient semi-supervised segmentation framework by integrating structured medical text priors with a dual-branch vision model.
Engineered a complementary dual-branch network combining a CNN (V-Net) for fine local features and a Vision Transformer (SwinUNETR) for long-range spatial context.
Integrated a task-aware attention module for voxel-level text-visual alignment alongside a discrepancy-aware mechanism targeting inter-branch prediction inconsistencies.
Evaluated segmentation accuracy across three public datasets (SegTHOR, MM-WHS, and MyoPS) under reduced annotation regimes.
Achieved a mean Dice score of 70.86% on SegTHOR using only 10% labeled data, exceeding the second-best method by 5.8%.
Yielded substantial performance gains on geometrically challenging anatomical structures, including the esophagus and right atrium.

Cite This Study

Chen et al. (2026) studied this question.

synapsesocial.com/papers/6a87bf6e91c33485a1fe7843 https://doi.org/10.1016/j.metrad.2026.100236

View Full Paper

Ask AI

Bookmark

July 1, 2026Meta-RadiologyOpen Access

Voxel-Level Text-Visual Alignment with Discrepancy-Aware Fusion for Semi-Supervised Multi-Organ Segmentation

View Full Paper

Ask AI

Bookmark

Authors

WCWenghao Chen SLSheng Lian JLJiayao Liu

Discussion

Member takes

Overview

Computational study demonstrates improved semi-supervised segmentation accuracy across multiple organs in CT and MRI scans, highlighting potential for annotation-efficient clinical workflows.

Key Points

To establish an annotation-efficient semi-supervised segmentation framework by integrating structured medical text priors with a dual-branch vision model.
Engineered a complementary dual-branch network combining a CNN (V-Net) for fine local features and a Vision Transformer (SwinUNETR) for long-range spatial context.
Integrated a task-aware attention module for voxel-level text-visual alignment alongside a discrepancy-aware mechanism targeting inter-branch prediction inconsistencies.
Evaluated segmentation accuracy across three public datasets (SegTHOR, MM-WHS, and MyoPS) under reduced annotation regimes.
Achieved a mean Dice score of 70.86% on SegTHOR using only 10% labeled data, exceeding the second-best method by 5.8%.
Yielded substantial performance gains on geometrically challenging anatomical structures, including the esophagus and right atrium.

Cite This Study

Chen et al. (2026) studied this question.

synapsesocial.com/papers/6a87bf6e91c33485a1fe7843 https://doi.org/10.1016/j.metrad.2026.100236

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark