# Scaling 3D Visual Grounding in Abdominal CT

arXiv: https://arxiv.org/abs/2610.04095 · HTML: https://mlsanity.com/paper/2610.04095/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 16 of its 25 references that are in the map.
- First posted on arXiv 2 Oct 2026.

## Builds on (most cited first, 16 of 16)

- [Modeling Context in Referring Expressions](https://mlsanity.com/paper/1608.00272/) (2016)
- [Shikra: Unleashing Multimodal LLM's Referential Dialogue Magic](https://mlsanity.com/paper/2306.15195/) (2023)
- [MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding](https://mlsanity.com/paper/2104.12763/) (2021)
- [TotalSegmentator: robust segmentation of 104 anatomical structures in CT images](https://mlsanity.com/paper/2208.05868/) (2022)
- [MAttNet: Modular Attention Network for Referring Expression Comprehension](https://mlsanity.com/paper/1801.08186/) (2018)
- [Making the Most of Text Semantics to Improve Biomedical Vision--Language Processing](https://mlsanity.com/paper/2204.09817/) (2022)
- [BiomedParse: a biomedical foundation model for image parsing of everything everywhere all at once](https://mlsanity.com/paper/2405.12971/) (2024)
- [Merlin: A Computed Tomography Vision-Language Foundation Model and Dataset](https://mlsanity.com/paper/2406.06512/) (2024)
- [Large-Vocabulary Segmentation for Medical Images with Text Prompts](https://mlsanity.com/paper/2312.17183/) (2023)
- [Pillar-0: A New Frontier for Radiology Foundation Models](https://mlsanity.com/paper/2511.17803/) (2025)
- [DeepLesion: Automated Deep Mining, Categorization and Detection of Significant Radiology Image Findings using Large-Scale Clinical Lesion Annotations](https://mlsanity.com/paper/1710.01766/) (2017)
- [MIMO: A medical vision language model with visual referring multimodal input and pixel grounding multimodal output](https://mlsanity.com/paper/2510.10011/) (2025)
- [MedMO: Grounding and Understanding Multimodal Large Language Model for Medical Images](https://mlsanity.com/paper/2602.06965/) (2026)
- [Vision-Language Modeling in PET/CT for Visual Grounding of Positive Findings](https://mlsanity.com/paper/2502.00528/) (2025)
- [MedVL-SAM2: A unified 3D medical vision-language model for multimodal reasoning and prompt-driven segmentation](https://mlsanity.com/paper/2601.09879/) (2026)
- [Opportunistic Promptable Segmentation: Leveraging Routine Radiological Annotations to Guide 3D CT Lesion Segmentation](https://mlsanity.com/paper/2602.00309/) (2026)

## Abstract (arXiv)

Visual grounding models can enhance radiology workflows by linking report findings to image regions. This is particularly valuable for 3D CT, where findings often occupy a tiny fraction of the volume. Training 3D grounding models requires large sets of paired phrases and regions, and building such datasets is expensive, requiring radiologists to annotate images by hand. We posit that this supervision is already created implicitly during routine reporting, as radiologists frequently place 2D annotations (e.g., distance measurement, arrows) on key images to make measurements and to support report interpretation. We introduce an automated pipeline that converts these routine clinical annotations into large-scale phrase-region supervision for 3D visual grounding. The pipeline links each annotation to the corresponding finding in the report through metadata matching, then uses a promptable 3D segmentation model to convert the 2D annotation into a volumetric mask. This produces phrase-mask-volume datasets without requiring additional radiologist annotation. Applied to a single institution's clinical picture archiving and communication system (PACS), our approach generated 105K phrase-mask-volume triplets from 59K abdominal CT exams. We also introduce two abdominal CT grounding benchmarks, LocusBench-Onc and LocusBench-ED, which comprise 240 oncology and 260 emergency-department radiologist-reviewed phrase-mask-volume triplets, respectively, with the latter spanning 13 distinct categories such as appendicitis, hematoma, and hernia. We further introduce LocusCT, a 3D visual grounding model trained on this dataset, which achieves hit rates of 0.725 on LocusBench-Onc and 0.773 on LocusBench-ED, substantially outperforming comparator models. These results show that routine PACS annotations are a scalable, previously unused source of supervision for 3D visual grounding.