Feasibility study of a single breath-hold, 3D mDIXON pulse sequence for late gadolinium enhancement imaging of ischemic scar

James R. J. Foley, Graham J. Fent, Pankaj Garg, David A. Broadbent, Laura E. Dobson, Pei G. Chew, Louise A. E. Brown, Peter P. Swoboda, Sven Plein, David M. Higgins, John P. Greenwood

Research output: Contribution to journalArticlepeer-review

11 Citations (Scopus)
5 Downloads (Pure)


Background: Late gadolinium enhancement (LGE) imaging is well validated for the diagnosis and quantification of myocardial infarction (MI). 2D LGE imaging involves multiple breath-holds for acquisition of short-axis slices to cover the left ventricle (LV). 3D LGE methods cover the LV in a single breath-hold; however, breath-hold duration is typically long with images susceptible to motion artifacts.

Purpose/Hypothesis: To assess a single breath-hold 3D mDIXON LGE pulse sequence for image quality and quantitation of MI.

Study Type: Prospective.

Population: Ninety- two patients with prior MI.

Field Strength/Sequence: 1.5T cardiac MRI protocol using both conventional 2D phase sensitive inversion recovery and 3D mDIXON LGE imaging 10 minutes following contrast administration in random order to avoid bias.

Assessment: Data were analyzed qualitatively for image quality (three observers). Quantitative assessment of myocardial scar mass (full-width half-maximum), scar transmurality, and contrast-to-noise ratio measurements were performed. Time for 2D and 3D LGE imaging was recorded.

Statistical Tests: Paired Student's t-test, Wilcoxon rank test, Cohen κ statistic, Pearson correlation, linear regression, and Bland–Altman analysis.

Results: Image quality scores were comparable between 3D and 2D LGE (1.4 ± 0.6 vs. 1.3 ± 0.5; P = 0.162). 3D LGE was associated with greater scar tissue mass (3D: 18.9 ± 17.5 g vs. 2D: 17.8 ± 16.2 g P = 0.03), although this difference was less pronounced when scar tissue was expressed as %LV mass (3D: 13.4 ± 9.9% vs. 2D: 12.7 ± 9.5% P = 0.07). For 3D vs. 2D scar mass there was a strong and significant positive correlation; Bland–Altman analysis showed mean mass bias of 1.1 g (95% confidence interval [CI]: –5.7 to 7.9). Segmental level agreement of scar transmurality between 3D and 2D LGE at the clinical viability threshold of 50% transmurality was excellent (κ = 0.870). 3D image acquisition (15.6 ± 1.4 sec) was just 5% of time required for 2D images (311.6 ± 43.2 sec) P < 0.0001.

Data Conclusion: Single breath-hold 3D mDIXON LGE imaging allows quantitative assessment of MI mass and transmurality, with comparable image quality, in vastly shorter overall acquisition time compared with standard 2D LGE imaging.
Original languageEnglish
Pages (from-to)1437-1445
Number of pages9
JournalJournal of Magnetic Resonance Imaging
Issue number5
Early online date31 Dec 2018
Publication statusPublished - May 2019

Cite this