Mitral regurgitation quantification by cardiac magnetic resonance imaging (MRI) remains reproducible between software solutions

Wellcome Open Research

article

Ciaran Grafton-Clarke¹  George Thornton²  Benjamin Fidock³  Gareth Archer³  Rod Hose³  Rob J. van der Geest⁴  Liang Zhong⁵  Andrew J. Swift³  James M. Wild³  Estefania De Gárate⁶  Chiara Bucciarelli-Ducci⁷  Sven Plein⁸  Thomas A. Treibel²  Marcus Flather¹  Vassilios S. Vassiliou¹  Pankaj Garg¹ 

[1] Medical School, University of East Anglia;Institute for Cardiovascular Sciences, University College London Hospitals NHS Trust;Department of Infection, University of Sheffield;Department of Radiology, Leiden University Medical Center;National Heart Centre, Duke NUS Graduate Medical School;British Heart Institute;Royal Brompton & Harefield NHS Trust, Guy's and St Thomas' NHS Foundation Trust;Leeds Institute of Cardiovascular and Metabolic Medicine, University of Leeds

关键词: Magnetic resonance imaging;    Mitral valve insufficiency;    Reproducibility of results.;   

DOI  :  10.12688/wellcomeopenres.17200.3

学科分类：内科医学

来源: Wellcome

【 摘 要 】

Background: The reproducibility of mitral regurgitation (MR) quantification by cardiovascular magnetic resonance (CMR) imaging using different software solutions remains unclear. This research aimed to investigate the reproducibility of MR quantification between two software solutions: MASS (version 2019 EXP, LUMC, Netherlands) and CAAS (version 5.2, Pie Medical Imaging).Methods: CMR data of 35 patients with MR (12 primary MR, 13 mitral valve repair/replacement, and ten secondary MR) was used. Four methods of MR volume quantification were studied, including two 4D-flow CMR methods (MRMVAV and MRJet) and two non-4D-flow techniques (MRStandard and MRLVRV). We conducted within-software and inter-software correlation and agreement analyses.Results: All methods demonstrated significant correlation between the two software solutions: MRStandard (r=0.92, p<0.001), MRLVRV (r=0.95, p<0.001), MRJet (r=0.86, p<0.001), and MRMVAV (r=0.91, p<0.001). Between CAAS and MASS, MRJet and MRMVAV, compared to each of the four methods, were the only methods not to be associated with significant bias.Conclusions: We conclude that 4D-flow CMR methods demonstrate equivalent reproducibility to non-4D-flow methods but greater levels of agreement between software solutions.

【 授权许可】

CC BY