Fast cardiac magnetic resonance (CMR) protocol for biventricular functional assessment and tissue characterisation.
Authors
Affiliations (7)
Affiliations (7)
- Perioperative Cardiology and Cardiovascular Imaging Department, Centro Cardiologico Monzino IRCCS, Milan, Italy.
- Department of Health Sciences (DISSAL), University of Genoa, Genoa, Italy.
- Department of Medical Sciences, University of Turin, Turin, Italy.
- University Cardiologic Unit, Interdisciplinary Department of Medicine, Polyclinic University Hospital, University of Bari Aldo Moro, Bari, Italy.
- GE HealthCare, Munich, Germany.
- Department of Radiology, Azienda Ospedaliero Universitaria, University of Cagliari, Cagliari, Italy.
- Perioperative Cardiology and Cardiovascular Imaging Department, Centro Cardiologico Monzino IRCCS, Milan, Italy; Department of Biomedical, Surgical and Dental Sciences, University of Milan, Milan, Italy. Electronic address: [email protected].
Abstract
Clinical use of cardiovascular magnetic resonance (CMR), reference tool for cardiac function and myocardial tissue assessment, is frequently limited by long acquisition times. This study aimed to compare conventional "standard protocol" (ASSET bSSFP cine plus 2D single-segment PSIR LGE) with novel "fast protocol" incorporating deep-learning reconstruction (Sonic DL bSSFP cine and 2D multisegment PSIR LGE with AIR Recon DL), focusing on image quality, functional measurements, myocardial characterisation, and overall scan duration. One-hundred consecutive patients with known or suspected myocardial disease underwent both protocols. Participants were predominantly male (78%), mean age 52 ± 16 years, mean BMI 25.0 ± 4.3 kg/m<sup>2</sup>. Clinical indications included arrhythmias (26%), hypertrophic cardiomyopathy (19%), and coronary artery disease (13%). Cine image quality was comparable between ASSET bSSFP and Sonic DL bSSFP (Likert score 4.59 vs 4.56, p = 0.682), with no differences in ventricular size, function, or left ventricle mass. However, Sonic DL cine markedly shortened acquisition time for long-axis and short-axis stacks (38 vs 61 s and 125 vs 227 s respectively, both p < 0.001). Similarly, 2D-MS PS LGE preserved diagnostic quality (Likert score 4.60 vs 4.51) while reducing acquisition time for long-axis and short-axis stacks (25 vs 77 s and 78 vs 302 s respectively, both p < 0.001). The "fast" protocol reduced total scan time by nearly 60%, enabling comprehensive CMR completion in under 10 min. A deep learning-based "fast" CMR protocol significantly reduces scan time without compromising volumetric accuracy or image quality, resulting a feasible option for routine clinical practice.