CMR-Multi基准挑战赛
大规模多中心心脏MRI基准数据集，提供4D Cine和LGE序列的全视图及专家级心脏结构注释，支持心脏数字孪生与AI模型开发。
基本信息
资源简介
CMR-Multi基准挑战赛
【数据集背景】
本数据集源站标题为《qutaiping/CMR_multi_baseline》。源站首次发布：2026-04-08。
【数据内容】
源站原始描述：Universal Multi-Sequence, Multi-Center and Multi-View CMR Segmentation Challenge With cardiovascular diseases (CVDs) remaining the leading global health threat, Cardiac Magnetic Resonance (CMR) imaging stands as the "gold standard" for diagnosis. Consequently, the accuracy and robustness of automated CMR analysis are paramount. While traditional CMR challenges have largely focused on single-sequence, single-slice, or isolated segmentation tasks, these limitations often hinder the generalizabilit…
【适用方向】
影像类数据可用于医学图像分析模型（检测、分割、分类）的训练与评测，也可用于标注质量与成像参数相关性的研究。
【获取与许可】
源站页面：https://github.com/qutaiping/CMR_multi_baseline
【使用注意】
· 源站最后修改于 2026-06-22，引用时请注意版本时效。
下载信息
注册下载
需要注册 Kaggle 账号并登录后下载，适合需要跟踪下载记录和使用 API 的用户。
暂未开放公开下载
无需注册即可直接获取公开样本或文档，适合快速预览和评估数据集质量。
免登录有偿下载
公开数据集受托下载与技术交付服务。
提供高速下载与技术交付服务（收技术服务费，非数据销售）
暂未开放当前数据集主要来源为 Kaggle 公开托管，完整影像包建议通过原始链接或 Kaggle API 获取。
使用方式
数据集获取
git clone https://github.com/qutaiping/CMR_multi_baseline.git
curl -L -o repo.zip https://github.com/qutaiping/CMR_multi_baseline/archive/refs/heads/main.zip
unzip repo.zip
源站 README 摘录（使用方式）
Universal Multi-Sequence, Multi-Center and Multi-View CMR Segmentation Challenge
With cardiovascular diseases (CVDs) remaining the leading global health threat, Cardiac Magnetic Resonance (CMR) imaging stands as the “gold standard” for diagnosis. Consequently, the accuracy and robustness of automated CMR analysis are paramount. While traditional CMR challenges have largely focused on single-sequence, single-slice, or isolated segmentation tasks, these limitations often hinder the generalizability of AI models in complex clinical scenarios.
To bridge this gap, the CMR-Multi Benchmark Challenge is jointly launched by top-tier international institutions, including the Beijing Academy of Artificial Intelligence (BAAI), Beijing Anzhen Hospital, Jinan University, the University of Auckland, and the University of Sydney. As a large-scale, multi-center challenge, it unifies 4D Cine (3D+t) and Late Gadolinium Enhancement (LGE) sequences for the first time. Covering full views of the Short Axis (SAX), 2-Chamber (2CH), and 4-Chamber (4CH), the dataset provides expert-level, precise annotations for ventricles, myocardium, atria, and pathological scars. This initiative aims to provide foundational data support for Cardiac Digital Twins and large-scale medical models, fostering the development of clinically interpretable, cross-platform general artificial intelligence technologies.
🎯 Challenge Theme: From Structural Segmentation to Functional Quantification
Hosted in conjunction with the 27th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2026) and the 1st Workshop on Medical World Models (MWM4MICCAI), this challenge focuses on the following core tasks:
Task 1: Cine Multi-sequence Segmentation & Wall Motion Abnormality Analysis
Automatic Segmentation: Utilize multi-sequence Cine data to achieve automatic segmentation of the Left/Right Ventricle (LV/RV), Myocardium (MYO), and Left/Right Atrium (LA/RA).
Functional Assessment: Evaluate regional myocardial motion through 4D sequences to quantify cardiac function, specifically the Left Ventricular Ejection Fraction (LVEF).
Task 2: LGE Multi-sequence Scar Segmentation & Quantitative Analysis
Pathological Segmentation: Perform segmentation of myocardial scars and atrial wall structures within LGE sequences.
Quantitative Metrics: Precisely quantify scar volume, transmur
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。
数据缺失？
依托陕西、四川两大基地，我们与超过十家三甲医院建立直接合作关系，覆盖合规授权、采集治理、专业标注、数据交付的全流程，为AI医疗团队提供即拿即用的高质量临床数据。
⚡️ 需要数据支持或标注服务？立即联系我们获取专业方案。
