How to participate to the challenge


To participate to ACDC challenge and get access to the involved datasets, you simply have to create an account through the dedicated online evaluation website (through the "Register" option presents at the top right of the webpage). Once you will register, you could access to the data of the challenge through your personal account.

Results submission guidelines for the segmentation part of the contest

  • Fully and semi automatic methods can compete for the segmentation of one particular structure (e.g. the myocardium) or the full set of structures of interest (left ventricular endocardium, myocardium and right ventricular endocardium) for both end diastole and end systole phases. Ranking of the segmentation methods will be performed for each structure of interest separately.

  • Participants are invited to submit their segmentation results through the dedicated online evaluation platform (using their own account) by uploading the corresponding files.

  • For each patient, the participant has to upload two files: one corresponding to the End Diastolic instant (ED) and one for the End Systolic (ES) instant.

  • Each file has to be named according to the following naming convention:

    • patientXXX_ED.nii.gz
    • patientXXX_ES.nii.gz

    where XXX should be replaced by the 3 digits number of the patient under segmentation (this number appears in the filename of the corresponding input data). For instance, for patient 1 you should upload the following files: patient001_ED.nii.gz and patient001_ES.nii.gz.

  • All results should be saved into a nifti image format. Each segmented image should involve discrete values using the following convention:

    • 0 -> background
    • 1 -> right ventricle
    • 2 -> myocardium
    • 3 -> left ventricle

  • Each segmented volume should be expressed in the corresponding input MR image space.

  • In order to optimize the computation of the different error measures, the participants are invited to respect the same properties as the ones of the reference segmented images whose information are provided here.

Please refer to this citation for any use of the ACDC database

  • 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