X射线报告数据集
高质量的匿名化X射线图像与放射报告配对数据集，用于胸部疾病多标签分类和自动报告生成。
基本信息
资源简介
本数据集包含高质量的匿名化X射线图像与放射科报告的配对数据，覆盖胸部病理（如肺炎、心脏肥大）等多标签分类任务，用于自动报告生成、医学图像分析及跨模态学习。数据模态为医学影像和文本，适用于训练和评估放射学AI模型。
下载信息
注册下载
Tips: 该数据集需要在对应的数据源网站注册通过后，才能进行数据下载，注册有对应要求，或者需要收费。
暂未开放公开下载
Tips: 该数据集属于公开下载，应该可以免费公开下载。
免登录有偿下载
Tips: 该数据集 Qianfanghub 可以协助提供有偿下载服务，注意，服务不针对数据相关产权，只是技术服务费。
提供高速下载与技术交付服务（收技术服务费，非数据销售）
暂未开放千方医数集，医疗数据集部分，是为社区服务的公开医疗数据集搜索引擎，并不存储或者下载原始的任何数据。 如果您有其他医疗数据需求，可以和客服联系，或者下工单。我们有强大的三甲医疗机构帮助您提供个性化的医疗数据定制、采集、标注服务。
使用方式
数据集获取（ModelScope）
方式一：MsDataset（Python）
# 前置依赖: pip install modelscope
from modelscope.msdatasets import MsDataset
ds = MsDataset.load("Kratos-AI/medical-imaging", subset_name="default", split="train")
print(ds)
方式二：CLI（命令行）
pip install modelscope
modelscope download dataset Kratos-AI/medical-imaging
数据集卡片摘录（源站）
X-ray Reports Dataset
This dataset contains high-quality (“A-grade”) anonymized X-ray images paired with radiology reports. It has been carefully curated, cleaned, and verified to ensure accuracy, completeness, and compliance with privacy standards (e.g., HIPAA/GDPR), making it suitable for high-stakes or research-grade model training.
Contact
For queries or collaborations related to this dataset, contact:
Supported Tasks
-
Task Categories:
- Image Classification
- Image-to-Text Generation
-
Supported Tasks:
- Radiology report generation from X-ray images
- Multi-label classification of thoracic pathologies (e.g., pneumonia, cardiomegaly)
- Medical image analysis for triage support
- Cross-modal learning for vision-language models
- Feature extraction for diagnostic AI research
Languages
- Primary Language: English (radiology reports)
Dataset Creation
Curation Rationale
This dataset was created to advance medical AI research by providing paired X-ray images and radiology reports for tasks like automated report generation and disease detection. It aims to support the development of robust, generalizable models for radiology.
Source Data
- Contributors: De-identified data from hospital archives and public medical repositories
- Collection Process: Images sourced from PACS systems (2015–2023), reports authored by board-certified radiologists, anonymized to remove patient identifiers.
Other Known Limitations
- Size: Limited to ~10,000 samples, which may restrict generalization
- Demographic Bias: Overrepresentation of adult urban patients; limited pediatric data
- Image Quality: Variations in X-ray resolution or equipment may affect consistency
- Label Noise: Potential errors in report-based labels extracted via NLP
Intended Uses
✅ Direct Use
- Training and benchmarking models for radiology report generation
- Research in medical image-to-text generation
- Development of AI tools for radiology triage and decision support
- Academic research in medical imaging and natural language processing
❌ Out-of-Scope Use
- Clinical diagnosis without human radiologist oversight
- Commercial use without proper attribution or ethical review
- Applications violating patient priv
许可
cc-by-4.0
数据加载示例（图像类）
from PIL import Image
import glob, os
files = (glob.glob(os.path.join(path, "**", "*.png"), recursive=True)
+ glob.glob(os.path.join(path, "**", "*.jpg"), recursive=True)
+ glob.glob(os.path.join(path, "**", "*.tif"), recursive=True))
print("图像文件数:", len(files))
img = Image.open(files[0]); print("尺寸/模式:", img.size, img.mode)
# torchvision Dataset 方式:
# from torchvision import datasets
# ds = datasets.ImageFolder(path) # 要求 子目录=类别
目录组织与标注格式以源站说明和下载后实际文件为准。
精度瓶颈？数据缺失？
当前公开数据无法满足您的算法精度？千方提供针对 其他 的高质量、多模态真实临床数据定制解决方案。
