Research map / Vision-language and multimodal / Human and embodied modeling
PETAR: Localized Findings Generation with Mask-Aware Vision-Language Modeling for PET Automated Reporting
Danyal Maqbool, Changhee Lee, Zachary Huemann, Samuel D. Church, Matthew E. Larson, Scott B. Perlman, Tomas A. Romero, Joshua D. Warner, Meghan Lubner, Xin Tie, Jameson Merkow, Junjie Hu and 2 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add 3D focal prompt (ablation) | PETAR-4B baseline without mask, CT, TS · PETARSeg-11K test (ablation) | GREEN: 0.071 -> 0.226 | Improved |
| add 3D focal prompt (ablation) | PETAR-4B baseline without mask, CT, TS · PETARSeg-11K test (ablation) | CIDEr: 0.132 -> 0.397 | Improved |
| add TotalSegmentator pretraining (ablation) | PETAR-4B (mask+CT+focal) · PETARSeg-11K test (ablation) | GREEN: 0.239 -> 0.257 | Improved |
| add TotalSegmentator pretraining (ablation) | PETAR-4B (mask+CT+focal) · PETARSeg-11K test (ablation) | RaTEScore: 0.613 -> 0.713 | Improved |
| add lesion mask input without focal prompt (ablation) | PETAR-4B baseline · PETARSeg-11K test (ablation) | BLEU: 0.485 -> 0.480 | No clear effect |
| add CT modality without focal prompt (ablation) | PETAR-4B baseline · PETARSeg-11K test (ablation) | GREEN: 0.071 -> 0.060 | Hurt |
| add mask input on top of focal prompt (ablation) | PETAR-4B (focal only) · PETARSeg-11K test (ablation) | CIDEr: 0.397 -> 0.381 | Mixed |
| add CT on top of focal prompt (ablation) | PETAR-4B (focal only) · PETARSeg-11K test (ablation) | RaTEScore: 0.698 -> 0.587 | Mixed |
| add mask to CT+focal (ablation) | PETAR-4B (CT+focal) · PETARSeg-11K test (ablation) | RaTEScore: 0.587 -> 0.613 | Improved |
| finetune on PETARSeg-11K | M3D-RAD · PETARSeg-11K test | GREEN: 0.003 -> 0.071 | Improved |
| finetune on PETARSeg-11K | Qwen3-VL-8B · PETARSeg-11K test | GREEN: 0.022 -> 0.066 | Improved |
| use PETAR-4B instead of finetuned M3D-RAD | PETAR-4B · PETARSeg-11K test | GREEN: 0.071 -> 0.257 | Improved |
Closest papers by meaning
- Toward a Vision-Language Foundation Model for Medical Data: Multimodal Dataset and Benchmarks for Vietnamese PET/CT Report Generation (NeurIPS 2025 · similarity 0.86)
- Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging (NeurIPS 2025 · similarity 0.85)
- Cross-Modal Interactive Perception Network with Mamba for Lung Tumor Segmentation in PET-CT Images (CVPR 2025 · similarity 0.83)
- Enhancing vision-language models for medical imaging: bridging the 3D gap with innovative slice selection (NeurIPS 2024 · similarity 0.83)
- Autoregressive Sequence Modeling for 3D Medical Image Representation (AAAI 2025 · similarity 0.83)
- Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding (ICLR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
