Instance-Guided Report Anchoring for Text-Free 3D Abnormality Segmentation in Chest CT

Researchers propose Instance-Guided Report Anchoring (IGRA), a model-agnostic module that improves 3D abnormality segmentation in chest CT by anchoring instance representations to radiology report findings during training, discarding text at inference. IGRA boosts Dice by 22.5% over the strongest image-only baseline (30.93 vs. 25.25) on ReXGroundingCT and matches VoxTell on single-finding subsets, with consistent gains across four backbones and zero-shot datasets.

arXiv:2609.00447v1 Announce Type: new Abstract: Accurate 3D abnormality segmentation in chest CT requires dense spatial supervision, but obtaining expert voxel-level labels is costly. Radiology reports, however, are routinely generated during clinical interpretation and contain instance-specific descriptions that can provide additional guidance without new dense annotation. Existing vision-language grounding methods typically require report-derived findings at inference, making localization dependent on paired text and limiting each forward pass to a queried finding. We propose Instance-Guided Report Anchoring IGRA , a model-agnostic module that preserves the correspondence between each annotated abnormality instance and the report finding that describes it. IGRA pools each instance representation and anchors it to the corresponding finding embedding during training; all text-related components are discarded at inference. We further reformulate free-text grounding on ReXGroundingCT as multi-label volumetric segmentation by merging same-category instances, allowing all abnormality categories to be predicted in one image-only forward pass. IGRA improves Dice by 22.5% over the strongest image-only baseline 30.93 vs. 25.25 and is comparable to VoxTell on the single-finding subset 30.29 vs. 30.43 . Applied unchanged to four standard 3D segmentation backbones, IGRA improves Dice and hit rate across all architectures. Zero-shot evaluation on LIDC-IDRI, PleThora, and a private in-house dataset further shows consistent gains over image-only baselines.