Evaluation of a highly under sampled, Dixon-based native 3D MR angiography in patients with aortopathies. (RF_FR_306)

MR angiography (MRA) is the imaging modality of choice for follow-up patients with aortopathies. Non-contrast sequences often have the disadvantage of long scan times resulting in limited image quality. Therefore, this study investigates a novel non-contrast MRA sequence (Dixon3D) and its diagnostic value in patients with aortopathies.

Methods:

This prospective study included 35 patients (31 M, 62 ± 15y) who underwent MRA exams to assess the thoracic aorta at 1.5T (MAGNETOM Sola, Siemens Healthineers AG). The indications were follow-up aneurysm in 21 patients (60%) and post aortic/valve surgery in 14 patients (40%). The research protocol included both a state-of-the-art, clinically approved bSSFP3D sequence with fat saturation and the new Dixon3D sequence based on a GRE readout with Dixon fat-water separation and compressed sensing acceleration. Acquisition parameters are listed in table 1. Reader 1 analysed all datasets qualitatively based on a 5-point Likert scale (1=non-diagnostic, 2=poor, 3=acceptable, 4=good, 5=excellent) and quantitatively with dedicated software (Terrarecon) at 6 predefined locations (proximal/distal ascending aorta; mid arch; proximal/mid descending aorta; hiatus). The signal intensity ratio (SIR) was measured as ROI in the aorta and ROI in a homogenous skeletal muscle. Statistical analysis included paired t-test and intraclass correlation coefficient. Reliability was evaluated with intra- and inter-reader (reader 2) agreement. P-values < 0.05 were considered statistically significant.
 


Results:

One dataset was excluded due to incomplete imaging data; therefore, the final analysis was performed on 34 data sets. The analysis was successfully conducted in all predefined vessel segments on both sequences. A total 408 segments were evaluated. The image quality of Dixon3D was superior to the bSSFP3D (3.9+/-0.9 vs. 3.5+/-1.0; p=0.02), figure 1A. The proximal and distal ascending aortic measurements were significantly different between both sequences (41.2±6.0 mm vs. 41.8±6.2 mm; mean 0.55 (95%CI 0.07-1.05); p=0.02; 37.4±4.3mm vs. 38.6±4.3mm; mean 1.21 mm (95%CI 0.69-1.72); p= < 0.0001). All other vessel segments showed no significant differences. The SIR was significantly (p < 0.0001) higher in the bSSFP3D (3.75±0.9) compared to Dixon3D (2.58±0.5). The intra- and inter-reader agreement was good to excellent, table 2. The scan time was significantly (p < 0.0001) shorter with Dixon3D (9.9±-2.9min) compared to bSSFP3D (18.5±4.9min), figure 1B.

Conclusion:

Dixon3D is a faster acquisition technique compared to the clinical standard bSSFP3D. Although Dixon3D has a lower SIR ratio the image quality is superior compared to bSSFP3D due to less image noise. Therefore, it may allow robust aortic measurements especially in the ascending aorta and with high reliability which may increase precision.