Evaluation of a Free-breathing Cardiac Cine-MRI Sequence With Image Reconstructions by Deep-Learning in Ischemic Heart Disease

Cardiac Magnetic Resonance Imaging, Deep-Learning, Left Ventricular Ejection Fraction, Magnetic Resonance Imaging

Magnetic resonance imaging, Cardiac magnetic resonance imaging, Deep-Learning, Left ventricular ejection fraction, Left ventricular end systolic volume, left ventricular end diastolic volume

Today, MRI is the gold standard for the precise assessment of left ventricular volume and function, but presents the drawback of having a long acquisition time and of generating motion artifacts, in particular respiratory artifacts, requiring repeated sequences in apnea to cover the whole cardiac volume. These apneas are difficult to achieve in patients with ischemic heart disease and may lead to degradation of the images, an increase in the duration of the examination by repeated acquisitions and therefore to diagnostic inaccuracies. Artificial intelligence, already used in practice in cardiac MRI for automatic segmentation of the heart chambers, improves radiological interpretation with rapid and precise measurements. Deep-learning, which is part of artificial intelligence, would allow the reconstruction of cine-MRI sequences in free breathing, in order to overcome the artifacts from respiratory motions, and the improvement of diagnostic performance while improving examination conditions for patients. Patients coming for a cardiac MRI for the assessment of ischemic heart disease will be eligible to the protocol. If the patient agrees to participate, a free-breathing cardiac cine-MRI sequence with Deep Learning based image reconstruction will be added to the usual protocol. No follow-up will be required in this study.

None listed

France