Gated vs Ungated Contrast Enhanced MR Angiography of the Thorax: A Paired Quantitative Comparison.

Background: Conventional 3D contrast enhanced MR angiographic (CEMRA) acquisitions of the heart and great vessels are performed without cardiac gating, introducing motion artifacts that degrade image quality, signal uniformity and blood-myocardial contrast. Gated acquisitions can mitigate these artifacts, but quantitative comparison between matched gated and ungated datasets is lacking due to the confounding effects of changes in the blood concentration of gadolinium agents, which does not apply during ferumoxytol enhanced imaging. We aim to evaluate signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) across major intrathoracic regions on gated vs ungated CEMRA following ferumoxytol enhancement.

Methods: The gated sequence employed compressed sensing (CS) image reconstruction with undersampling factors up to 12X and a diastolic acquisition window of 350 ms. Resolution and bandwidth were similar between gated and ungated sequences. Thirty-one adult patients referred for clinical indications underwent paired gated and ungated, breath-held CEMRA during the steady state distribution of ferumoxytol (4 mg/kg). Signal intensities were measured in the inferior vena cava (IVC), superior vena cava (SVC), ascending aorta (AA), and left and right ventricular chambers. Background noise was defined as the mean SD across three background ROIs. Interventricular septum (IVS) signal was the mean of three septal ROIs. SNR was defined as mean signal divided by background SD. CNR was the blood-pool–septum difference divided by background SD. Paired comparisons used normality testing followed by paired t or Wilcoxon tests, with effect sizes and 95% CIs (MedCalc, version 23.3.7).

Results: For SNR, the gated sequence outperformed the ungated sequence in every region. Mean±SD values were 194.4±61.9 vs 82.0±39.7 in the IVC, 225.8±70.4 vs 87.5±34.7 in the SVC, 208.3±62.8 vs 84.4±28.9 in the AA, 175.5±49.6 vs 75.1±26.2 in the LV blood pool, and 188.1±62.7 vs 80.7±31.6 in the RV blood pool. The corresponding paired differences (gated minus CEMRA) were 112.44 (95% CI 89.40–135.47, p< 0.0001), 138.29 (95% CI 112.22–164.35, p< 0.0001), 123.94 (95% CI 101.18–146.71, p< 0.0001), 100.32 (95% CI 81.98–118.67, p< 0.0001), and 107.46 (95% CI 84.56–130.36, p< 0.0001), respectively. For CNR, gated imaging likewise exceeded CEMRA: LV 152.7±46.6 vs 58.7±21.6 with a paired difference of 93.95 (95% CI 75.19–112.71, p< 0.0001) and RV 166.2±61.1 vs 63.7±27.0 with a paired difference of 102.55 (95% CI 78.78–126.33, p< 0.0001). Image quality in regions normally subject to cardiac motion was consistently superior on gated MRA and image quality in other regions was similar on both sequences.

Conclusion: In paired patient data across major cardiac ROIs, gated MRI provides consistently higher SNR and CNR than ungated steady-state imaging. These findings support routine adoption of gating when evaluation of structures that undergo significant motion during the cardiac cycle is required.

Figure 1: 25-year-old female patient with history of Ehlers-Danlos being monitored for developing aortic aneurysm. Aortic and cardiac motion artifact present on ungated images (b,d) is eliminated on the corresponding gated images (a,c).
Figure 2: 21-year-old male patient with D-Transposition of the Great Arteries. Note the improvement in definition of the aortic valve leaflets (arrows) and ventricular chambers (arrowheads) on gated (b,d) vs ungated (a,c) images.
Figure 3: 49-year-old female patient with ascending aortic aneurysm. Note the improvement in definition of the aortic lumen (arrows) and ventricular chambers (arrowheads) on gated (b,d) vs ungated (a,c) images.