Accelerating non-contrast MR angiography of the thoracic aorta using compressed SENSE with deep learning reconstruction

Jan P. Janssen, Roman J. Gertz, Juliana Tristram, Marvin A. Spurek, Kenan Kaya, Robert Terzis, Robert Hahnfeldt, Thorsten Gietzen, David Maintz, Thorsten Persigehl, Kilian Weiss, Lenhard Pennig, Carsten Gietzen,
Accelerating non-contrast MR angiography of the thoracic aorta using compressed SENSE with deep learning reconstruction,
European Journal of Radiology,
Volume 192,
2025,
112403,
ISSN 0720-048X,
https://doi.org/10.1016/j.ejrad.2025.112403.
(https://www.sciencedirect.com/science/article/pii/S0720048X25004899)
Abstract: Purpose
REACT (Relaxation-Enhanced Angiography without ContrasT) is a reliable non-contrast magnetic resonance angiography for imaging of the thoracic aorta but remains time-consuming. This study evaluates acceleration of image acquisition using compressed sensing and parallel imaging (Compressed SENSE, CS) combined with deep learning-based image reconstruction (CS-AI).
Methods
In this prospective single-center study, 40 volunteers underwent ECG- and navigator-triggered 3D REACT at 3 T using CS acceleration factor 4 (CS4; reference standard) and 8 (CS8). CS8 data were reconstructed with standard and CS-AI methods (CS8-AI). Two radiologists measured aortic diameters, rated subjective image quality and performed pairwise comparisons. Additionally, objective image quality metrics were calculated.
Results
Median scan time was reduced by 44 % (CS4: 8:41 min; CS8/CS8-AI: 4:52 min). All techniques showed excellent agreement in aortic diameter measurements (mean differences < 0.2 mm; P > 0.999). CS8-AI demonstrated reduced mean absolute deviation from CS4 compared to CS8 (0.67 vs. 0.77 mm; P = 0.003), and measurement variance was 40–50 % lower with CS8-AI than with CS8 (inter-/intrarater: P < 0.001), and comparable to CS4. CS8 showed significantly lower subjective image quality scores than CS4 (3.70[3.33–4.00] vs. 4.25[3.90–4.50]; P < 0.001), while CS8-AI showed comparable or higher scores (4.40[4.00–4.70]; P = 0.076). Forced-choice comparisons favored CS4 over CS8 (90 % vs. 2.5 %; P < 0.001), but no preference was observed between CS4 and CS8-AI (42.5 % vs. 37.5 %; P > 0.999). Objective metrics predominantly confirmed the subjective results.
Conclusion
Deep learning-based reconstruction enables the acquisition of REACT of the thoracic aorta in less than five minutes while preserving high image quality and maintaining excellent measurement reproducibility.
Keywords: Thoracic aorta; Magnetic resonance angiography; Non-contrast-enhanced magnetic resonance angiography; Compressed sensing; Deep Learning