---
license: cc-by-nc-sa-4.0
task_categories:
- image-segmentation
language:
- es
- en
tags:
- medical
- mri
- cardiac
- heart
- cardiac-mri
- cine-mri
- segmentation
- acdc
pretty_name: ACDC-Automated Cardiac Diagnosis Challenge
size_categories:
- 1K<n<10K
---

![Screenshot 2023-06-11 at 23.19.31.png](https://s3.amazonaws.com/moonup/production/uploads/6226bae1c8655fec3995a41d/cO9OKcYBO7-MbDZJopF6J.png)

General information

The overall ACDC dataset was created from real clinical exams acquired at the University Hospital of Dijon. Acquired data were fully anonymized and handled within the regulations set by the local ethical committee of the Hospital of Dijon (France). Our dataset covers several well-defined pathologies with enough cases to (1) properly train machine learning methods and (2) clearly assess the variations of the main physiological parameters obtained from cine-MRI (in particular diastolic volume and ejection fraction). The dataset is composed of 150 exams (all from different patients) divided into 5 evenly distributed subgroups (4 pathological plus 1 healthy subject groups) as described below. Furthermore, each patient comes with the following additional information : weight, height, as well as the diastolic and systolic phase instants.

Tasks

The main task of this dataset is the semantic segmentation of the heart in cardiac magnetic resonance images, specifically the endocardium and myocardium. The present task is very relevant for the detection of cardiovascular diseases. Segmentation is a very time-consuming process, so automatically performing the segmentation with Artificial Intelligence algorithms can be extremely beneficial to reduce the time spent in a manual segmentation. In this way, a very relevant bottleneck can be avoided and cardiovascular diseases can be detected in a timely manner.

## License & Attribution

Released under [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/),
the same license applied by the original ACDC organizers. Cite the original
creators (see Reference below).

**Changes made in this mirror:** the original ACDC NIfTI release is re-hosted with
its native `training/` and `testing/` per-patient layout (each `patientNNN/` folder
retains its `Info.cfg`, `_4d.nii.gz` cine volume, end-diastole/end-systole
`_frameNN.nii.gz` images and `_gt.nii.gz` ground-truth masks) plus a `train.jsonl`
index for convenient loading. **No voxel intensities or segmentation labels were
altered** — the volumes and masks are the unmodified ACDC data.

## Reference

O. Bernard, A. Lalande, C. Zotti, F. Cervenansky, et al.
"Deep Learning Techniques for Automatic MRI Cardiac Multi-structures Segmentation and Diagnosis: Is the Problem Solved ?" in IEEE Transactions on Medical Imaging, vol. 37, no. 11, pp. 2514-2525, Nov. 2018
doi: 10.1109/TMI.2018.2837502