mlsanityData to 14 Oct 2026

CheXGround: Anatomical Region Tokens for Grounded Longitudinal Chest X-ray Interpretation

arXiv 2608.30758 ↗First posted 31 Aug 2026cs.CVcited by 0

On the map: LLM reasoning → Clinical and biomedical applications

also in 1 more
Abstract (arXiv)

Recent radiology multi-modal language models have made substantial progress in chest X-ray report generation, visual question answering, and temporal reasoning. While longitudinal chest X-ray interpretation compares sequential examinations to describe change, visual grounding aims to connect clinical language with localized image evidence. Although longitudinal modeling and visual grounding have each advanced radiology language models, how localized visual evidence can support longitudinal interpretation remains under-explored. We introduce CheXGround, a region-grounded longitudinal chest X-ray language model that represents paired studies through corresponding anatomical regions. CheXGround extracts anatomical regions from current and prior radiographs, encodes them as temporally enhanced Region-of-Interest (ROI) tokens, and combines them with global temporal image context during generation. To connect these region tokens with clinical text, we propose Temporal Region--Phrase Alignment, a pretraining objective that aligns temporal anatomical representations with localized report phrases. We evaluate CheXGround on single-study and longitudinal Visual Question Answering (VQA), longitudinal findings generation, temporal grounded VQA, and anatomical grounding. Across these tasks, CheXGround improves clinical language quality, temporal reasoning, and localization accuracy over recent baselines. Our results suggest that organizing longitudinal evidence at the anatomical level is a strong representation for grounded radiology language modeling. Project page: https://adonaydem.github.io/chexground-website

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Clinical and biomedical applications, one of 17 directions in LLM reasoning. Builds on 57 of its 75 references that are in the map. First posted on arXiv 31 Aug 2026.

Builds on

57 of 75 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024MedThink: Explaining Medical Visual Question Answering via Multimodal Decision-Making Rationale18 Apr

Other references · 56

  1. 2017Attention Is All You Needcited by 60,361
  2. 2017Mask R-CNNcited by 7,277
  3. 2020Deformable DETR: Deformable Transformers for End-to-End Object Detectioncited by 3,422
  4. 2025Qwen3-VL Technical Reportcited by 2,793
  5. 2023LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Daycited by 1,305
  6. 2023LISA: Reasoning Segmentation via Large Language Modelcited by 776
  7. 2020Contrastive Learning of Medical Visual Representations from Paired Images and Textcited by 641
  8. 2023MiniGPT-v2: large language model as a unified interface for vision-language multi-task learningcited by 625
  9. 2019MIMIC-CXR-JPG, a large publicly available database of labeled chest radiographscited by 597
  10. 2020PathVQA: 30000+ Questions for Medical Visual Question Answeringcited by 475
  11. 2023GLaMM: Pixel Grounding Large Multimodal Modelcited by 455
  12. 2021SLAKE: A Semantically-Labeled Knowledge-Enhanced Dataset for Medical Visual Question Answeringcited by 449
38 more
  1. 2020Generating Radiology Reports via Memory-driven Transformercited by 368
  2. 2022Making the Most of Text Semantics to Improve Biomedical Vision--Language Processingcited by 339
  3. 2020CheXbert: Combining Automatic Labelers and Expert Annotations for Accurate Radiology Report Labeling Using BERTcited by 337
  4. 2021RadGraph: Extracting Clinical Entities and Relations from Radiology Reportscited by 319
  5. 2023MEDITRON-70B: Scaling Medical Pretraining for Large Language Modelscited by 274
  6. 2017On the Automatic Generation of Medical Imaging Reportscited by 216
  7. 2023Learning to Exploit Temporal Structure for Biomedical Vision-Language Processingcited by 211
  8. 2022Cross-modal Memory Networks for Radiology Report Generationcited by 185
  9. 2021Exploring and Distilling Posterior and Prior Knowledge for Radiology Report Generationcited by 162
  10. 2022Multi-Granularity Cross-modal Alignment for Generalized Medical Visual Representation Learningcited by 159
  11. 2024MAIRA-2: Grounded Radiology Report Generationcited by 145
  12. 2018TieNet: Text-Image Embedding Network for Common Thorax Disease Classification and Reporting in Chest X-rayscited by 139
  13. 2024Exploring scalable medical image encoders beyond text supervisioncited by 132
  14. 2024Groma: Localized Visual Tokenization for Grounding Multimodal Large Language Modelscited by 119
  15. 2019Knowledge-driven Encode, Retrieve, Paraphrase for Medical Image Report Generationcited by 102
  16. 2024One Token to Seg Them All: Language Instructed Reasoning Segmentation in Videoscited by 100
  17. 2021Chest ImaGenome Dataset for Clinical Reasoningcited by 90
  18. 2021Joint Learning of Localized Representations from Medical Images and Reportscited by 81
  19. 2023MAIRA-1: A specialised large multimodal model for radiology report generationcited by 74
  20. 2023EHRXQA: A Multi-Modal Question Answering Dataset for Electronic Health Records with Chest X-ray Imagescited by 53
  21. 2024VideoGLaMM: A Large Multimodal Model for Pixel-Level Visual Grounding in Videoscited by 51
  22. 2026MedGemma 1.5 Technical Reportcited by 45
  23. 2024MedVersa: A Generalist Foundation Model for Medical Image Interpretationcited by 38
  24. 2023Utilizing Longitudinal Chest X-Rays and Reports to Pre-Fill Radiology Reportscited by 35
  25. 2024PadChest-GR: A Bilingual Chest X-ray Dataset for Grounded Radiology Report Generationcited by 35
  26. 2024HERGen: Elevating Radiology Report Generation with Longitudinal Datacited by 33
  27. 2022Self-supervised vision-language pretraining for Medical visual question answeringcited by 30
  28. 2024Libra: Leveraging Temporal Images for Biomedical Radiology Analysiscited by 29
  29. 2024VividMed: Vision Language Model with Versatile Visual Grounding for Medicinecited by 28
  30. 2025SpaceVLLM: Endowing Multimodal Large Language Model with Spatio-Temporal Video Grounding Capabilitycited by 28
  31. 2024Artemis: Towards Referential Understanding in Complex Videoscited by 25
  32. 2025RadVLM: A Multitask Conversational Vision-Language Model for Radiologycited by 25
  33. 2023Controllable Chest X-Ray Report Generation from Longitudinal Representationscited by 23
  34. 2022CheXRelNet: An Anatomy-Aware Model for Tracking Longitudinal Relationships between Chest X-Rayscited by 19
  35. 2024Anatomical Structure-Guided Medical Vision-Language Pre-trainingcited by 18
  36. 2024M4CXR: Exploring Multi-task Potentials of Multi-modal Large Language Models for Chest X-ray Interpretationcited by 17
  37. 2025DDaTR: Dynamic Difference-aware Temporal Residual Network for Longitudinal Radiology Report Generationcited by 8
  38. 2022UnICLAM:Contrastive Representation Learning with Adversarial Masking for Unified and Interpretable Medical Vision Question Answeringcited by 6

50 most cited of 56 All 56 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026CheXTemporal: A Dataset for Temporally-Grounded Reasoning in Chest Radiography
  2. 2026AnatomiX, an Anatomy-Aware Grounded Multimodal Large Language Model for Chest X-Ray Interpretation
  3. 2026TRACE: Temporal Radiology with Anatomical Change Explanation for Grounded X-ray Report Generation
  4. 2025CoCa-CXR: Contrastive Captioners Learn Strong Temporal Structures for Chest X-Ray Vision-Language Understanding
  5. 2026GRCD: Grounded Region Change Detection for Multi-Finding Chest X-Ray Pairs
  6. 2025VICCA: Visual Interpretation and Comprehension of Chest X-ray Anomalies in Generated Report Without Human Feedback
  7. 2026MI-CXR: A Benchmark for Longitudinal Reasoning over Multi-Interval Chest X-rays
  8. 2025XBench: A Comprehensive Benchmark for Visual-Language Explanations in Chest Radiography