In total, 200 cases from the DETERMINE (Defibrillators to Reduce Risk by Magnetic Resonance Imaging Evaluation) cohort are available for this challenge. All the data are in de-identified DICOM format following the HIPAA regulations. Permission to use the data in this challenge has been granted by the DETERMINE study, subject to the CAP-STACOM11 Terms of Data Usage Agreement. Informed consent and local IRB approval have been obtained.

The DETERMINE cohort consists of patients suffering from myocardial infarction and impaired LV contraction. Each case includes short-axis and long-axis views of cine-MRI. These images were acquired with Steady-State Free Precession (SSFP) MR imaging protocol with typical thickness ≤ 10mm, gap ≤ 2mm, TR 30-50ms, TE 1.6ms, flip angle 60 deg, FOV 360mm and 256x256mm image matrix. Patients were asked to hold their breath for 8-15 sec. during MR acquisition session. The studies were acquired at multiple sites using a range of different scanner types and manufacturers.

[A screenshot of the CAP DICOM web-viewer.]

Test set

The test set consists of 100 randomly selected cases, each image slice is accompanied by the corresponding ground truth images. Participants may use the test set to develop, to train or to test their segmentation algorithm.

[download the test set (size: 2.71 GB)]

If you have difficulties to download the whole test set file, we broke it down into the following smaller 10 zip files:

  1. Test set 01 ..................... [download] (size: 200 MB)
  2. Test set 02 ..................... [download] (size: 296 MB)
  3. Test set 03 ..................... [download] (size: 546 MB)
  4. Test set 04 ..................... [download] (size: 284 MB)
  5. Test set 05 ..................... [download] (size: 294 MB)
  6. Test set 06 ..................... [download] (size: 201 MB)
  7. Test set 07 ..................... [download] (size: 258 MB)
  8. Test set 08 ..................... [download] (size: 168 MB)
  9. Test set 09 ..................... [download] (size: 260 MB)
  10. Test set 10 ..................... [download] (size: 201 MB)

Validation sets

The validation sets consists of 5 sub-groups, each having 20 cases. All sub-groups only contain MRI data from short-axis and long-axis views. No ground truth images are given.

All participants must at least submit their segmentation results from one sub-group. When submitting the segmentation results, please start from the lowest sub-group ID. There are no preferences when we grouped the validation set. All sub-groups were created by random selection.

  1. Validation set sub-group 01 ............................... [download] (size: 409.1 MB)
  2. Validation set sub-group 02 ............................... [download] (size: 447.8 MB)
  3. Validation set sub-group 03 ............................... [download] (size: 530.6 MB)
  4. Validation set sub-group 04 ............................... [download] (size: 783.5 MB)
  5. Validation set sub-group 05 ............................... [download] (size: 460.1 MB)

If you have experience any trouble to download the test or validation sets, please contact us.

Browsing the data set

You can browse the test and validation sets by using the CAP DICOM web-viewer as shown in figure above. Click the following link and use the same account information to login: [access to CAP-DB].

Note: if you download DICOM images directly from the CAP database, you will not get the same filename as in the ZIP files here.