Sunnybrook Cardiac Data
Sunnybrook Cardiac Data (SCD), MICCAI 2009 Cardiac MR Left Ventricle Segmentation Challenge
Short-axis cine cardiac MRI of 45 patients from Sunnybrook Health Sciences Centre, Toronto (healthy, LV hypertrophy, heart failure with and without infarction), with expert endocardial and epicardial LV contours at end-diastole and end-systole. CC0.
Overview
The Sunnybrook Cardiac Data (SCD) is a set of short-axis cine cardiac MRI exams with manual left ventricular contours, contributed by Perry Radau and colleagues at Sunnybrook Health Sciences Centre in Toronto. It was built for the Cardiac MR Left Ventricle Segmentation Challenge at a MICCAI 2009 workshop and is now hosted by the Cardiac Atlas Project under a CC0 public domain dedication. It is used to develop and compare left ventricle segmentation methods and to measure ejection fraction and myocardial mass.
Composition
There are 45 patients, one cine exam each, in four groups: 9 healthy, 12 with left ventricular hypertrophy, 12 with heart failure without infarction and 12 with heart failure after infarction. Heart failure groups have an ejection fraction below 40%, split by the presence of late gadolinium enhancement; the hypertrophy group has a normal ejection fraction and an indexed LV mass above 83 g/m2. The patient spreadsheet lists 32 men and 13 women aged 23 to 88. For the 2009 challenge the exams were split at random into three sets of 15 (training, testing and on-line contest). The download comes as five DICOM batches plus the contours, LV models and the patient spreadsheet, which maps the Cardiac Atlas Project ids to the original challenge ids.
Acquisition
Exams were collected in clinical routine on a 1.5 T GE Signa scanner. Short-axis SSFP cine images were acquired in breath-holds of 10 to 15 seconds with 20 phases over the cardiac cycle. Six to twelve slices cover the heart from the atrioventricular ring to the apex, with 8 mm slice thickness, an 8 mm gap, a 320 mm field of view and a 256 x 256 matrix. Images are given without preprocessing after reconstruction.
Annotations
An experienced cardiologist drew endocardial and epicardial contours of the left ventricle on all slices at end-diastole and end-systole, with papillary muscles and trabeculations counted as cavity. A second cardiologist confirmed every contour.
Known limitations
- Single centre, single scanner and a small cohort.
- Only the left ventricle is contoured, and only at two phases.
- Thick slices with large gaps limit through-plane resolution.
- Group sizes are set by design and do not reflect clinical prevalence.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 45 subjects.
Sex
- Male 32 71%
- Female 13 29%
Age
mean 61.1± 15.2 · range 23 to 88Contrast / sequence
subjects, values can overlap
- Cine MRI 45 100%
Condition
subjects, values can overlap
- Heart failure 24 53%
- Left ventricular hypertrophy 12 27%
- Myocardial infarction 12 27%
- Healthy control 9 20%
Scanner vendor
subjects
- GE HealthCare 45 100%
Field strength
subjects
- 1.5 T 45 100%
Country
subjects
- Canada 45 100%
Split
subjects
- Training 15 33%
- Validation 15 33%
- Test 15 33%
License and access
Our reading of the license, not legal advice. Before you use the data, read the original license and confirm that your use is allowed. We take no responsibility for how you use a dataset. Full disclaimer
Download without an account
Creative Commons Zero 1.0 Universal
Public domain dedication. Do anything with the data, including commercial use, without asking and without having to give credit.
What you can do
- Yes
- Yes
- Yes
- Yes
What you can share
- Yes
- Yes
- Yes
What you must do
- No
- Share alike No
- No
- No
- Manuscript review No
- Release code No
- Return results No
- Delete after use No
Limits
- No
- Location limits No
Citation
Radau P, Lu Y, Connelly K, Paul G, Dick AJ, Wright GA. Evaluation Framework for Algorithms Segmenting Short Axis Cardiac MRI. The MIDAS Journal - Cardiac MR Left Ventricle Segmentation Challenge (2009). http://hdl.handle.net/10380/3070
All numbers
Every number on this page, as stored in stats.csv, with its source.
| Measure | Breakdown | Value | Source |
|---|---|---|---|
| Subjects | total 45 cine MRI exams | 45 | cap-scd |
| Subjects | contrast=cine Short-axis SSFP cine | 45 | radau2009 p. 5 |
| Subjects | field_strength=1.5 GE Signa 1.5 T | 45 | radau2009 p. 5 |
| Subjects | vendor=ge | 45 | radau2009 p. 5 |
| Subjects | country=CA Sunnybrook Health Sciences Centre | 45 | cap-scd |
| Subjects | condition=healthy N group: EF above 55% and no hypertrophy | 9 | cap-scd Pathology group table |
| Subjects | condition=left_ventricular_hypertrophy HYP group: EF above 55% and LV mass index above 83 g/m2 | 12 | cap-scd Pathology group table |
| Subjects | condition=heart_failure Sum of the disjoint HF (12) and HF-I (12) groups, both with EF below 40% | 24 | cap-scd Pathology group table |
| Subjects | condition=myocardial_infarction HF-I group: heart failure with late gadolinium enhancement | 12 | cap-scd Pathology group table |
| Subjects | split=train Original 2009 challenge split | 15 | radau2009 p. 5 |
| Subjects | split=validation Original 2009 challenge testing set | 15 | radau2009 p. 5 |
| Subjects | split=test Original 2009 challenge on-line contest set | 15 | radau2009 p. 5 |
| Subjects | sex=male | 32 | cap-scd-patientdata Gender |
| Subjects | sex=female | 13 | cap-scd-patientdata Gender |
| Subjects | age=20-29 | 2 | cap-scd-patientdata Age |
| Subjects | age=30-39 | 1 | cap-scd-patientdata Age |
| Subjects | age=40-49 | 8 | cap-scd-patientdata Age |
| Subjects | age=50-59 | 8 | cap-scd-patientdata Age |
| Subjects | age=60-69 | 11 | cap-scd-patientdata Age |
| Subjects | age=70-79 | 10 | cap-scd-patientdata Age |
| Subjects | age=80-89 | 5 | cap-scd-patientdata Age |
| Mean age | total | 61.1 | cap-scd-patientdata Age |
| Age SD | total Sample standard deviation | 15.2 | cap-scd-patientdata Age |
| Median age | total | 61 | cap-scd-patientdata Age |
| Minimum age | total | 23 | cap-scd-patientdata Age |
| Maximum age | total | 88 | cap-scd-patientdata Age |
Sources
The keys used in the table above.
- cap-scd Cardiac Atlas Project, Sunnybrook Cardiac Data page website
- radau2009 Radau et al. 2009, The MIDAS Journal (challenge website pages, MICCAI 2009 LV Segmentation Challenge) paper
- cap-scd-patientdata SCD patient data spreadsheet (scd_patientdata.csv, CC0), counted from the Gender and Age columns computed