The local Institutional Review Board approved the study protocol (approval number: 59_21B), which complies with the Health Insurance Portability and Accountability Act criteria and the Declaration of Helsinki. All patients signed an informed consent form before study enrolment.

We retrospectively screened patients who underwent aortic CT examinations with third-generation dual-source CT (Somatom Force; Siemens Healthcare GmbH, Forchheim, Germany) over 15 months (January 2020 to May 2021) for participation in this study, identifying 168 patients. The exclusion criteria were death (n = 3) or debilitating illness (n = 21), age < 18 years (n = 3), colonization with multi-drug resistant microbes (n = 4), residential distance > 100 km from the hospital (n = 6), metallic aortic foreign materials (n = 34), or contraindications for MRI (i.e., claustrophobia or unsafe implants; n = 12). We contacted the remaining 85 patients by telephone and letter, of which 25 agreed to participate in this study. We did not ask for the reasons for non-participation to respect the patient’s privacy.

The study population comprised eight women and 17 men (mean age: 62 ± 13 years). Initial indications for CT were suspected aortic dissection (n = 14), aortic aneurysm (n = 5), and workup before planned interventions, such as transcatheter aortic valve implantation or percutaneous coronary intervention (n = 3), unknown structure at the aortic root (n = 1), known aortic dissection (n = 1), and aortic coarctation (n = 1).

At study entry, the participants’ mean weight was 90 ± 22 kg, mean height was 174 ± 11 cm, and mean body mass index was 30 ± 6 kg/m2. Their mean heart rate during the MRI exam was 68 ± 8 beats/min.

Cardiovascular MRI was performed on a 3-T MRI system (MAGNETOM Vida; Siemens Healthcare, Erlangen, Germany) with dedicated phased-array receiver coils (an 18-channel body coil array and 32-channel spine coil array). A single coil array was used for the thoracic aorta, and when clinically indicated, a second coil array was added to assess the abdominal aorta.

The prototypical native MRA acquisition was planned as a coronal slab. The number of slices was adapted individually to include the entire aorta and heart. Geometrical parameters included a field of view of 450 × 450 × 156 ± 24 mm3 and an acquired and reconstructed resolution of 1.2 × 1.2 × 1.2 mm3. A two-point Dixon with echo times of 1.3 or 2.9 ms was used. An undersampling factor of R = 11 was achieved using a Poisson-disc-like incoherent sampling pattern combined with an iterative compressed sensing reconstruction approach. An adiabatic T2-preparation pulse was used for optimal vessel contrast. Prospective ECG-gating in an end-diastolic phase was used to avoid heart motion. A cross-beam navigator was placed onto the liver dome, and the position of the liver-diaphragm interface was tracked before every acquisition window to mitigate artifacts from respiratory motion. An accept-reject algorithm was applied with an acceptance window of ± 4 mm at end-expiration. All images were reconstructed inline in the MRI system.

CTA imaging was performed on a third-generation dual-source scanner (SOMATOM Force; Siemens Healthcare, Erlangen, Germany) covering the entire thorax and abdomen. The scans were prospectively ECG triggered in the high-pitch mode (collimation = 2 × 192 × 0.6 mm, pitch = 3.2), and data were only acquired during end-diastole to avoid heart motion. For each patient, an arterial phase acquisition with bolus triggering in the ascending aorta was performed in the head-feet direction. When clinically indicated, an additional venous phase acquisition in the head-feet direction was performed 20 s after the first pass. Automatic tube current modulation (Care Dose 4D; reference = 180–204 mAs, depending on patient girth) and automatic tube voltage selection (Care kV; reference = 120 kV) were used for all patients. A total of 50–80 mL (mean = 64.7 mL ± 11 mL) of iodinated contrast agent (Imeron 350, Bracco, Milan, Italy), followed by 50 mL of 0.9% NaCl, were administered through a peripheral indwelling cannula at the right or left upper extremity at a 4 mL/s flow rate. Images were reconstructed in 0.6 mm slices in the transversal plane using smooth (Bv40) kernels using iterative reconstructions.

A dedicated vascular image interpretation workflow supported the evaluation of the aortic diameters (CT Vascular; Syngo.via VB50; Siemens Healthcare GmbH, Erlangen, Germany) with automated centerline definitions. Manual corrections of the centerlines were added by the readers when necessary. The effective inner diameter and cross-sectional area were documented at predefined aortic levels (i.e., aortic annulus, sinus of Valsalva, sinotubular junction, mid ascending aorta, proximal aortic arch, mid aortic arch, proximal descending aorta, mid descending aorta, diaphragm level, celiac trunk level, and above the bifurcation) [2].

Two board-certified reviewers ([blinded for review], with 10 and 7 years of experience in cardiovascular imaging, respectively), who were blinded to all clinical and imaging data, evaluated the datasets. The presentation started in the original orientation, and images were reviewed in free multiplanar reformation angulations at the discretion of the raters using a dedicated three-dimensional (3D) viewer (Horos v. 3.3.6; distributed under the LGPL license by Horosproject.org).

Overall image quality and water-fat separation were rated on a five-point scale [23]: 5 = excellent image quality, interpretable with no artifacts; 4 = good image quality, interpretable with minimal artifacts; 3 = average image quality, interpretation mildly degraded by image artifacts; 2 = below average image quality, interpretable but moderately degraded; 1 = poor image quality, uninterpretable images. The image quality of the myocardium, the aorta at the 11 abovementioned predefined levels, the thoracic and abdominal side branches, the pulmonary arteries, and the vena cavae were evaluated on a similar scale (Table 1). In cases of impaired image quality, the reason was documented in each case.

Interval-level data were evaluated for normality using the Shapiro–Wilk test. Data are presented as mean ± standard deviation (SD) or median (range). Data were compared using a paired t-test or Wilcoxon’s signed-rank test. A p-value < 0.05 was considered statistically significant. Bland–Altman plots and box plots were analyzed. Inter-rater agreement was evaluated using Cohen’s kappa value (κ), with κ interpreted as follows [24]: 0 < κ ≤ 0.2 = slight agreement, 0.2 < κ ≤ 0.4 = fair agreement, 0.4 < κ ≤ 0.6 = moderate agreement, 0.6 < κ ≤ 0.8 = substantial agreement, 0.8 < κ < 1.0 = almost perfect agreement, and κ = 1.0 as perfect agreement. All statistical analysis was performed using MedCalc Statistical software (version 20.218; MedCalc Software Ltd, Ostend, Belgium).