LUNA25-MedSAM2
收藏资源简介：
LUNA25-MedSAM2数据集是一个医学图像数据集，包含了6163个经过MedSAM2标注的肺结节。该数据集首先使用MedSAM2 Lesion CT分割模型对每个病变进行点提示，然后手动检查分割结果并修订所需的掩码。该数据集可用于训练和开发医学图像分割模型。
The LUNA25-MedSAM2 dataset is a medical imaging dataset consisting of 6163 lung nodules annotated with MedSAM2. Specifically, the MedSAM2 Lesion CT segmentation model was first utilized to generate point prompts for each lesion, followed by manual inspection of the segmentation outputs and revision of the corresponding masks as needed. This dataset can be used for training and developing medical image segmentation models.
LUNA25-MedSAM2 数据集概述
基本信息
- 语言: 英文 (en)
- 标签: 医学 (medical)
- 许可证: CC-BY-NC-4.0
- 任务类别: 图像分割 (image-segmentation)
数据集描述
- 该数据集包含来自 LUNA25 挑战赛 的 6163 个肺结节标注。
- 标注过程分为两步：
- 使用 MedSAM2 Lesion CT 分割模型对每个病变进行点提示分割。
- 人工检查分割结果并修正（约 880 个结节经过人工修正）。
作者信息
- 主要作者: Jun Ma, Zongxin Yang (共同第一作者), Bo Wang (通讯作者)
- 机构:
- 多伦多大学健康网络 AI 协作中心
- 向量人工智能研究所
- 哈佛医学院生物医学信息学系
- 多伦多大学健康网络 Peter Munk 心脏中心
- 多伦多大学计算机科学系
- 多伦多大学实验室医学与病理生物学系
- Roche Canada 和 Genentech
使用方式
python
安装所需包
pip install datasets
加载数据集
from datasets import load_dataset
下载并加载数据集
dataset = load_dataset("wanglab/LUNA25-MedSAM2")
访问训练集
train_dataset = dataset["train"]
显示第一个示例
print(train_dataset[0])
引用要求
使用该数据集时需同时引用 LUNA25 和 MedSAM2： bash @misc{LUNA25-Image, doi = {10.5281/ZENODO.14223624}, url = {https://zenodo.org/doi/10.5281/zenodo.14223624}, author = {Peeters, Dré and Obreja, Bogdan and Antonissen, Noa and Jacobs, Colin}, keywords = {Lung Cancer, Artificial Intelligence, Computed Tomography, Radiologists, Computer-Aided Diagnosis}, title = {The LUNA25 Challenge: Public Training and Development set - Imaging Data}, publisher = {Zenodo}, year = {2025}, copyright = {Creative Commons Attribution 4.0 International} }
@misc{LUNA25-Annotation, doi = {10.5281/ZENODO.14673658}, url = {https://zenodo.org/doi/10.5281/zenodo.14673658}, author = {Peeters, Dré and Obreja, Bogdan and Antonissen, Noa and Jacobs, Colin}, keywords = {Lung Cancer, Artificial Intelligence, Computed Tomography, Radiologists, Computer-Aided Diagnosis}, title = {The LUNA25 Challenge: Public Training and Development set - Annotation Data}, publisher = {Zenodo}, year = {2025}, copyright = {Creative Commons Attribution Non Commercial 4.0 International} }
@article{MedSAM2, title={MedSAM2: Segment Anything in 3D Medical Images and Videos}, author={Ma, Jun and Yang, Zongxin and Kim, Sumin and Chen, Bihui and Baharoon, Mohammed and Fallahpour, Adibvafa and Asakereh, Reza and Lyu, Hongwei and Wang, Bo}, journal={arXiv preprint arXiv:2504.63609}, year={2025} }
