The overall database was created from real clinical exams acquired at the University Hospital of Dijon (France).
A cardiologist with 10 years of experience in cardiology and MRI and a biophysicist with 20 years of experience of cardiovascular MRI have supervised the data acquisition.
Acquired data were fully anonymized and handled within the regulations set by the local ethical committee. As the data were collected retrospectively, and as the data are completely untraceable (because using the NifTI format, we discard all the administrative information included in the header), for the french law, and for the staff of the ethical committee of the University Hospital of Dijon, it was not necessary to do the process to have ethical approval number.
The database is made available to participants through two datasets (a training dataset and a testing dataset) after a personal registration under CC BY-NC-SA 4.0 license (Attribution-NonCommercial-ShareAlike).
The CC BY-NC-SA 4.0 license lets participants remix, adapt, and build upon the EMIDEC datasets non-commercially, as long as they credit the EMIDEC datasets and license their new creations under the identical terms.
The database includes:
For each case, there is a text file with the clinical information, a nifti file with the images and a nifti file with the labelled masks of each area (background (0), cavity(1), normal myocardium (2), myocardial infarction (3) and no-reflow (4)). Then the whole myocardium is the union of the masks 2, 3 and 4 The images and the masks are registered according to the center of gravity of the myocardium.
To view nifti files, use for example ITK-SNAP.
The acquisitions were obtained using Siemens MRI scanners (Area (1.5 T) and Skyra (3T)) during Conventional cardiovascular exam with no specific protocol. Short-axis slices of the DE-MRI have been extracted in a retrospective study.
In details, all acquisitions are ECG-gated, taken during breath-hold and performed 10 minutes after the injection of a gadolinium-based contrast agent. A T1-weighted Phase Sensitive Inversion Recovery (PSIR) sequence is used. Resulting MR images consist in a stack of short-axis slices from base to apex of the left ventricle with the following features: pixel spacing between 1.25 × 1.25 mm2 and 2 × 2 mm2, slice thickness of 8 mm and distance between slices between 8 and 13 mm (this information is indicated in the text file, and the Nifti file takes into account the distance between 2 slices). The variation of these parameters at the acquisition on 1.5 T or 3 T magnets allows us to deal with images with different signal to noise ratio. To prevent the drawback of the displacement of the heart location between slices due to different breath-holds, the slices are realigned according to the gravity center of the area defined by the epicardial contour. The raw input images will be provided using Nifti format, i.e. one file for the whole images covering the left ventricle, and one file with the ground-truths.
Moreover, clinical information will be provided in a text file: sex, age, tobacco (Y/N/former smoker), overweight (BMI > 25), arterial hypertension (Y/N), diabetes (Y/N), familial history of coronary artery disease (Y/N), ECG (ST+ (STEMI) or not), troponin (value), Killip max (between 1 and 4), ejection fraction of the left ventricle from echography (value), NTproBNP (value).