Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.04130 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911484323299328 |
|---|---|
| author | Zhang, Zichun Nie, Weizhi Guo, Honglin Su, Yuting |
| author_facet | Zhang, Zichun Nie, Weizhi Guo, Honglin Su, Yuting |
| contents | Counterfactual generation for chest X-rays (CXR) aims to simulate plausible pathological changes while preserving patient-specific anatomy. However, diffusion-based editing methods often suffer from structural drift, where stable anatomical semantics propagate globally through attention and distort non-target regions, and unstable pathology expression, since subtle and localized lesions induce weak and noisy conditioning signals. We present an inference-time attention regulation framework for reliable counterfactual CXR synthesis. An anatomy-aware attention regularization module gates self-attention and anatomy-token cross-attention with organ masks, confining structural interactions to anatomical ROIs and reducing unintended distortions. A pathology-guided module enhances pathology-token cross-attention within target lung regions during early denoising and performs lightweight latent corrections driven by an attention-concentration energy, enabling controllable lesion localization and extent. Extensive evaluations on CXR datasets show improved anatomical consistency and more precise, controllable pathological edits compared with standard diffusion editing, supporting localized counterfactual analysis and data augmentation for downstream tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_04130 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Mask-Guided Attention Regulation for Anatomically Consistent Counterfactual CXR Synthesis Zhang, Zichun Nie, Weizhi Guo, Honglin Su, Yuting Computer Vision and Pattern Recognition Counterfactual generation for chest X-rays (CXR) aims to simulate plausible pathological changes while preserving patient-specific anatomy. However, diffusion-based editing methods often suffer from structural drift, where stable anatomical semantics propagate globally through attention and distort non-target regions, and unstable pathology expression, since subtle and localized lesions induce weak and noisy conditioning signals. We present an inference-time attention regulation framework for reliable counterfactual CXR synthesis. An anatomy-aware attention regularization module gates self-attention and anatomy-token cross-attention with organ masks, confining structural interactions to anatomical ROIs and reducing unintended distortions. A pathology-guided module enhances pathology-token cross-attention within target lung regions during early denoising and performs lightweight latent corrections driven by an attention-concentration energy, enabling controllable lesion localization and extent. Extensive evaluations on CXR datasets show improved anatomical consistency and more precise, controllable pathological edits compared with standard diffusion editing, supporting localized counterfactual analysis and data augmentation for downstream tasks. |
| title | Mask-Guided Attention Regulation for Anatomically Consistent Counterfactual CXR Synthesis |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.04130 |