PETAR：面向 PET 自动报告的掩码感知视觉语言建模与局部化发现生成
PETAR: Localized Findings Generation with Mask-Aware Vision-Language Modeling for PET Automated Reporting
把 3D 病灶分割掩码作为显式空间条件注入视觉语言模型，并用 3D focal prompt 聚焦占体积极小（<0.1%）的病灶，这一「掩码感知 + 局部提示」思路可迁移到任何需要空间定位依据的医学或细粒度视觉语言生成任务。
原文依据 using a 3D focal prompt to capture fine-grained details of lesions that normally comprise less than 0.1% of the volume
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论为三维正电子发射断层成像（PET）生成自动化报告是医学影像中一项重要且具有挑战性的任务。PET 在肿瘤学中起着至关重要的作用，但由于全身三维体数据的复杂性、潜在临床发现种类繁多以及标注数据集稀缺，报告生成的自动化十分困难。为应对这些挑战，我们提出了 PETARSeg-11K，这是首个大规模、公开可用的数据集，提供三维 PET/CT 体数据与自由文本放射学发现之间的病灶级对应关系。它包含 11,356 条病灶描述，并与三维分割配对。其次，我们提出了 PETAR-4B，一个面向掩码感知、具空间定位能力的 PET/CT 报告生成的三维视觉语言模型。PETAR-4B 联合编码 PET、CT 与三维病灶分割掩码，使用三维焦点提示词来捕捉通常占体积极小（不足 0.1%）的病灶的细粒度细节。基于自动化指标的评估显示，PETAR-4B 显著优于所有二维与三维基线方法。一项由五位医师参与的人工研究——此类研究在自动化 PET 报告领域尚属首次——证实了该模型的临床实用性，并建立了自动化指标与专家判断之间的相关性。本工作提供了一个基础性数据集与一种新颖架构，推动了 PET 中三维医学视觉语言理解的发展。项目链接：https://github.com/DanyalMaq/petar-release/
英文原文摘要 1439 字符
Generating automated reports for 3D positron emission tomography (PET) is an important and challenging task in medical imaging. PET plays a vital role in oncology, but automating report generation is difficult due to the complexity of whole-body 3D volumes, the wide range of potential clinical findings, and the limited availability of annotated datasets. To address these challenges, we introduce PETARSeg-11K, the first large-scale, publicly available dataset that provides lesion-level correspondence between 3D PET/CT volumes and free-text radiological findings. It comprises 11,356 lesion descriptions paired with 3D segmentations. Second, we propose PETAR-4B, a 3D vision-language model designed for mask-aware, spatially grounded PET/CT reporting. PETAR-4B jointly encodes PET, CT, and 3D lesion segmentation masks, using a 3D focal prompt to capture fine-grained details of lesions that normally comprise less than 0.1% of the volume. Evaluations using automated metrics show PETAR-4B substantially outperforming all 2D and 3D baselines. A human study involving five physicians---the first of its kind for automated PET reporting---confirms the model's clinical utility and establishes correlations between automated metrics and expert judgment. This work provides a foundational dataset and a novel architecture, advancing 3D medical vision-language understanding in PET. Project link: https://github.com/DanyalMaq/petar-release/