PanTS
收藏资源简介:
PanTS是一个大规模、多机构合作的胰腺肿瘤分割数据集,旨在推动胰腺CT分析的深入研究。该数据集包含来自145个医疗中心的36,390张CT扫描图像,并由专家验证,对超过993,000个解剖结构进行了逐体素注释,包括胰腺肿瘤、胰腺头部、体部和尾部以及24个周围解剖结构,如血管/骨骼结构和腹部/胸部器官。每个扫描图像都包括元数据,如患者年龄、性别、诊断、对比阶段、平面间距、切片厚度等。在PanTS上训练的AI模型在胰腺肿瘤检测、定位和分割方面表现显著优于在现有公开数据集上训练的模型。PanTS作为同类中最大和最全面的资源,为开发和评估胰腺CT分析中的AI模型提供了一个新的基准。
PanTS is a large-scale, multi-institutional collaborative pancreatic tumor segmentation dataset designed to advance in-depth research on pancreatic CT analysis. This dataset comprises 36,390 CT scan images from 145 medical centers, which have undergone expert validation, with voxel-level annotations for more than 993,000 anatomical structures, including pancreatic tumors, the head, body and tail of the pancreas, and 24 surrounding anatomical structures such as vascular/bony structures and abdominal/chest organs. Each scan is accompanied by metadata including patient age, gender, diagnosis, contrast phase, planar spacing, slice thickness, and other relevant information. AI models trained on PanTS significantly outperform those trained on existing public datasets in pancreatic tumor detection, localization and segmentation. As the largest and most comprehensive resource of its kind, PanTS serves as a new benchmark for developing and evaluating AI models for pancreatic CT analysis.

- 1PanTS: The Pancreatic Tumor Segmentation Dataset约翰斯·霍普金斯大学计算机科学系 · 2025年



