CARE-Cardiac

Unified Cardiac Image Segmentation

Motivation

Cardiac diagnosis involves multiple interrelated tasks, such as delineating anatomical structures (e.g., atria, ventricles, vessels) and identifying pathological regions (e.g., scar) across diverse imaging modalities and acquisition protocols. Relying on separate models for each task leads to fragmented workflows, limited knowledge transfer, and poor adaptability to missing or heterogeneous data. A unified cardiac segmentation model addresses these limitations by enabling unified modeling across tasks.

Task

Develop a unified model capable of segmenting single cardiac structure (left atria), whole heart structures and structure and pathology (left ventricle, myocardium and scar).

*Note*: Participants are also encouraged to leverage external data.

Data

Multi-center and multi-modality datasets are provided for three sub-tasks. Please register here to access the data.

Training Dataset

1). Left Atrial Segmentation

Center Num. cases Modality
A 130 LGE MRI

2). Whole Heart Segmentation

Center Num. cases Modalities
A 20 CT
B 20 CT
C & D 20 MRI
E 26 MRI

3). Myocardial Pathology Segmentation

Center Num. cases Modality
A 81 LGE MRI
B 50 LGE MRI
C 45 LGE MRI
E 7 LGE MRI
F 9 LGE MRI
G 8 LGE MRI

Validation Dataset

1). Left Atrial Segmentation

Center Num. cases Sequences
A 10 LGE MRI
C 10 LGE MRI

2). Whole Heart Segmentation

Center Num. cases Modalities
A 20 CT
B 10 CT
C & D 20 MRI

3). Myocardial Pathology Segmentation

Center Num. cases Modality
D 25 LGE MRI

Test Dataset

1). Left Atrial Segmentation

Center Num. cases Modality
A 14 LGE MRI
B 20 LGE MRI
C 10 LGE MRI

2). Whole Heart Segmentation

Center Num. cases Modalities
A 20 CT
B 14 CT
C & D 20 MRI
F 16 MRI
G 20 CT

3). Myocardial Pathology Segmentation

Center Num. cases Modality
D 25 LGE MRI
B 16 LGE MRI

Metrics

Both in-distribution (i.e., seen centers) and out-of-distribution (OOD) (i.e., unseen centers) results of the three tasks will be reported in the leaderboard.

Task In-distribution Out-of-distribution
Single Cardiac Structure Lb1 Lb2
Whole Heart Structures Lb3 Lb4
Structure and Pathology Lb5 Lb6

Registration

To access the dataset, please register here.

Leaderboards

Single Cardiac Structure (ID)

Team Dice↑ HD(mm)↓
Med-AIR 0.9390 13.7725
Space 0.9392 15.8789
Cemrg 0.9230 73.7738
Xcare2025 0.8708 68.5828
Qipeng Su 0.7777 112.5363

Single Cardiac Structure (OOD)

Team Dice↑ HD(mm)↓
Qipeng Su 0.7732 28.2984
Space 0.7777 37.0773
Med-AIR 0.7768 37.4824
Xcare2025 0.7059 49.7781
Cemrg 0.4442 69.7284

Whole Heart Structures (ID)

Team Dice↑ HD(mm)↓
Med-AIR 0.9088 13.6256
Space 0.8994 14.6366
Xcare2025 0.8486 28.3754
Qipeng Su 0.8454 31.7669
Cemrg 0.8509 38.2703

Whole Heart Structures (OOD)

Team Dice↑ HD(mm)↓
Med-AIR 0.9261 10.4423
Cemrg 0.9038 17.0919
Space 0.8179 26.9377
Qipeng Su 0.8134 30.0940
Xcare2025 0.6187 63.3101

Structure and Pathology (ID)

Team Dice↑ HD(mm)↓
Med-AIR 0.6709 16.9396
Space 0.6705 17.7413
Qipeng Su 0.6096 21.1745
Xcare2025 0.5483 25.9150
Cemrg 0.5394 83.7736

Structure and Pathology (OOD)

Team Dice↑ HD(mm)↓
Med-AIR 0.7613 12.6665
Space 0.7558 12.7343
Cemrg 0.7236 14.3194
Qipeng Su 0.6474 18.4937
Xcare2025 0.6417 20.3698

Citations

Please cite these papers when you use the data for publications:

 @article{zhuang2019multivariate,
    title={Multivariate mixture model for myocardial segmentation combining multi-source images},
    author={Zhuang, Xiahai},
    journal={IEEE transactions on pattern analysis and machine intelligence},
    volume={41},
    number={12},
    pages={2933--2946},
    year={2019},
}

@article{zhuang2016multi,
  title={Multi-scale patch and multi-modality atlases for whole heart segmentation of MRI},
  author={Zhuang, Xiahai and Shen, Juan},
  journal={Medical image analysis},
  volume={31},
  pages={77--87},
  year={2016},
  publisher={Elsevier}
}

@article{li2022atrialjsqnet,
  title={AtrialJSQnet: a new framework for joint segmentation and quantification of left atrium and scars incorporating spatial and shape information},
  author={Li, Lei and Zimmer, Veronika A and Schnabel, Julia A and Zhuang, Xiahai},
  journal={Medical image analysis},
  volume={76},
  pages={102303},
  year={2022},
  publisher={Elsevier}
}

@article{GAO2023BayeSeg,
  title = {BayeSeg: Bayesian modeling for medical image segmentation with interpretable generalizability},
  journal = {Medical Image Analysis},
  volume = {89},
  pages = {102889},
  year = {2023},
  author = {Shangqi Gao and Hangqi Zhou and Yibo Gao and Xiahai Zhuang},
}