Acdc article
MRI cardiac segmentation has been a glaring medical imaging issue for the past decade. A quick search on pubmed reveals that literally thousands of papers have been published on that topic over the past few of years. With this challenge, we will provide the medical imaging community with the largest fully-annotated public MRI cardiac dataset ever made. As such, the richness of our dataset as well as its tight bound to every-day clinical issues has the potential of redefining the topic of computerized cardiac analysis and reset the counters of this research area. Furthermore, with the rise of deep learning methods applied to medical imaging, there has been a growing appetite for large and well-annotated datasets.

Our challenge has also a larger scope than previous cardiac challenges as it allows for two kinds of results: participants will be invited to submit their image segmentation results AND/OR their pathology prediction for each patient. Also, our dataset contains groundtruth data for the right ventricle, the endocardial and epicardial walls of the left ventricle which is unprecedented to our knowledge.

Finally, we provide an online evaluation platform to automatically compute the different metrics involved in each of the contest. By this way, all the results produced during the challenge will be computed from the same code (which will be made publicly available from the beginning of the challenge) and in a fair and reproducible manner. This will allow consistent evaluation of the solutions obtained by the proposed algortihm while highlighting the best performing ones, thus contributing to a faster clinical translation of groundbreaking technical advances.

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