Overview

NEWS

21st of February 2023
Part of the training dataset is available

31st of March 2023
The full training dataset will be available

30th of April 2023
The full testing dataset will be available
Goal of the MYOSAIQ challenge

Late gadolinium enhanced magnetic resonance images (LGE) are used in clinical practice to quantify the areas of myocardial muscle affected following a myocardial infarction (MI). In this context, the aim of this challenge is twofold:
  • compare the performance of automatic methods on the segmentation of the left ventricular endocardium and epicardium on LGE images;

  • compare the performance of automatic methods for the quantification of MI lesion size, at different phases of the longitudinal evolution of the disease on images acquired at the acute phase (4-8 days post-MI & reperfusion therapy) and delayed phase (1 month / 12 months post-MI & reperfusion).

illustration of the segmentation task in 3D MRI



Originality of the MYOSAIQ challenge

The originality of this challenge lies in the analysis of a large database composed of 467 MR-LGE exams and integrating the main various issues encountered in daily clinical cardiac imaging, namely:
  • the longitudinal evolution of the patho-physiology of the heart;

  • the presence of image artifacts;

  • the use of different MR acquisition techniques (3D versus 2D PSIR);

  • the use of different MR acquisition settings from different centers.