Approval for this single-centre study was granted by the local institutional review board. In light of the retrospective nature of the study, the institutional review board waived the requirement for written informed consent (reference number: blinded for submission).

Patient population

The authors reviewed the institutional image database at a tertiary care university hospital for visceral MRA studies between January 2021 and July 2023. Patients were included in the study if they had undergone a standardized protocol for the assessment of the arterial abdominal vasculature at 3 T, which included both REACT and 4D CE-MRA. The exclusion criteria were the absence or technical failure of any of the MRA sequences.

The following data were obtained from the medical records or observed during magnetic resonance imaging (MRI): age, sex, body mass index, indication for MRA, risk factors for atherosclerosis, and ascites.

Magnetic resonance imaging

A commercially available 3 T MRI system (Philips Ingenia, Philips Healthcare, Best, The Netherlands) with a standard 28-channel body coil without additional modifications was utilized. The abdominal scan protocol comprised T2-weighted turbo spin echo sequences in both the coronal and axial planes, REACT, 4D CE-MRA, and T1-weighted contrast-enhanced gradient-echo sequences with mDIXON for fat suppression in coronal and axial orientations.

For non-CE-MRA, 3D isotropic flow-independent REACT was acquired in the coronal plane covering the entire abdominal aorta from the diaphragm to the common iliac arteries. The REACT sequence comprises of T2 preparation and IR prepulses (which enhance the native blood signal with long T1 and T2) and a water and fat selective Dixon reconstruction based on a 7-peak fat model (mDIXON XD, Philips Healthcare), resulting in suppressed signal from adjacent background and fat [16, 25]. To account for respiratory motion, diaphragmatic pencil beam navigation was employed. The navigator was positioned on the dome of the right hemidiaphragm, and a 6 mm gating window was employed during end-expiration. Furthermore, the coronal image acquisition was combined with sagittal excitation (ENCASE, Philips Healthcare) to enable a reduced field of view in the left–right direction and to mitigate the impact of moiré artifacts [26]. Immediate image reconstruction was conducted using the standard hardware provided by the manufacturer. Compressed SENSE [27] (Philips Healthcare), which combines compressed sensing [28] and parallel imaging using SENSitivity Encoding (SENSE) [29], was used for acceleration of image acquisition. A variable-density incoherent sampling pattern was employed, whereby high-density sampling was conducted in the k-space centre and continuously increasing undersampling was applied towards the k-space periphery for the purpose of data acquisition. The data consistency and image sparsity were ensured by means of iterative L1 norm minimization for image reconstruction. Furthermore, the reconstruction was regularized through the use of SENSE parallel imaging and coil sensitivity distribution. An acceleration factor of 10 was employed, resulting in a nominal scan time of 2 min and 54 s.

For 4D CE-MRA, a 3D spoiled gradient echo sequence was employed. First, a native image was acquired as a mask. Afterwards, Gadobutrol (Gadovist, Bayer HealthCare Pharmaceuticals, Berlin, Germany; 0.1 ml/kg body weight)) was administered into an antecubital vein at a flow rate of 2 mL/second, followed by a 30 mL saline flush. Without triggering, acquisition in coronal plane was initiated after arrival of the contrast agent in the abdominal aorta, as determined by a bolus tracking sequence. Patients were instructed to maintain an end-expiratory breath hold during data acquisition. To achieve high spatiotemporal resolution, the acquisition was combined with SENSE (factor 6) and a keyhole technique with 20% of the central k-space data acquired in each dynamic (4D TRACK, Philips Healthcare) [30], resulting in a nominal scan time of 00:54 min. During reconstruction, the keyhole data were combined with outer k-space data from a reference scan.

Table 1 provides an overview of the imaging parameters of the MRA sequences.

Table 1 Imaging parameters of REACT relaxation-enhanced angiography without contrast and triggering and 4D CE-MRA contrast-enhanced magnetic resonance angiographyImage analysis

Two readers with three (R1) and six (R2) years of experience in MRA used a commercially available image viewer (DeepUnity Diagnost 1.1.1.1; Dedalus Healthcare Group, Bonn, Germany) to conduct independent reviews of the MRA images during separate sessions and in a randomized order. The readers were free to modify the window level and blinded to clinical and patient data. A four-week interval was maintained between the evaluation of the REACT and 4D CE-MRA datasets to minimize the potential for recall bias.

Assessment of subjectiv image quality

Based on vessel delineation, signal intensity, and contrast to adjacent tissue, the readers rated the vessel quality of the MRA datasets using a 4-point Likert scale (1: non-diagnostic, 2: poor, 3: fair, 4: excellent). The following arterial vessels were analyzed:

1.

Suprarenal abdominal aorta (SRA)

2.

Infrarenal abdominal aorta (IRA)

3.

Celiac trunk (CT)

4.

Superior mesenteric artery (SMA)

5.

Splenic artery (SA)

6.

Right renal artery (RRA)

7.

Left renal artery (LRA)

8.

Common hepatic artery (CHA)

9.

Proper hepatic artery (PHA)

10.

Gastroduodenal artery (GDA)

11.

Left gastric artery (LGA)

12.

Right hepatic artery (RHA)

13.

Left hepatic artery (LHA)

14.

Inferior mesenteric artery (IMA)

Assessment of visceral artery patency

In order to perform a subjective assessment of visceral vessel stenosis, MRA datasets were graded for stenosis affecting any of the aforementioned vessels using a 1–5 grading scale: Grade 1: regular patency, grade 2: stenosis, < 50% of vessel lumen, grade 3: stenosis, 50%-69% of vessel lumen, grade 4: stenosis, ≥ 70–99% of vessel lumen, grade 5: vessel occlusion. In the event of multiple stenoses, the most severe lesion was deemed the diagnostic grade and subjected to further analysis.

Assessment of vascular variants and other vascular findings

Furthermore, readers were advised to evaluate potential anatomical variants of the visceral arteries, including variants of hepatic arterial anatomy, aberrant renal arteries, and direct origin of the visceral arteries from the abdominal aorta. Moreover, readers were directed to assess MRAs for additional vascular findings beyond stenosis including vascular dissection and aneurysms of the aortic branches or visceral arteries.

Statistical analysis

The statistical analysis was conducted using IBM SPSS Statistics software (version 25.0, Armonk, NY, USA). The Shapiro–Wilk test was employed to ascertain a normal distribution. Categorical variables are presented as frequencies and corresponding percentages. Quantitative variables are presented as mean and standard deviation. Subjective ratings are presented as the median and interquartile range. Comparisons of normally distributed quantitative data were conducted using the Student's t-test, while non-normally distributed ordinal scaled data were compared using the Wilcoxon signed-rank test. A p-value of less than 0.05 was considered statistically significant for two-tailed tests.

Sensitivity and specificity of REACT regarding the detection of stenoses and other vascular findings were calculated considering 4D CE-MRA as a reference standard [7]. Cohen's kappa was employed to evaluate interrater and intersequence concordance in the identification of stenosis and other vascular findings. The interpretation of the degree of agreement was as follows: 0.01–0.2 slight, 0.21–0.4 fair, 0.41–0.6 moderate, 0.61–0.8 substantial, and 0.81–0.99 almost perfect.