The study population included n = 9 Landrace pigs selected from an already published study cohort from our group, in which dobutamine stress testing was performed [16, 23–25]. The experimental protocols were approved by the local bioethics committee of Berlin, Germany (G0138/17) and conform to the “European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes” (Council of Europe No 123, Strasbourg 1985). The range in weight of the pigs was 50.3 ± 9.2, the mean body surface area (BSA) was 2.4 ± 0.4 and no significant difference was observed between the groups.

The HR at rest was on average 104 ± 15 beats per minute (BPM), while during dobutamine infusion it reached on average 147 ± 11 BPM.

Before being transferred to the cardiac magnetic resonance (CMR) facility, female Landrace pigs were sedated and intubated. Ventilation was support with an MRI-compatible machine (Titus, Dräger Medical, Germany) with the following standardized settings: FiO2 of 0.5, I: E-ratio of 1:1.5, positive end-expiratory pressure of 5 mmHg, and a tidal volume of 10 ml/Kg. Moreover, the respiratory rate was adjusted, when required, to maintain an end-expiratory carbon dioxide partial pressure of 35–45 mmHg. Anaesthesia was kept stable with a combination of isoflurane, fentanyl, midazolam, ketamine and pancuronium. Dobutamine infusion was titrated, aiming at a 25% heart rate (HR) increase compared to baseline values. This protocol was established by our group in a small pilot study in which titration of dobutamine was assessed by left ventricle invasive conductance measurements. In one of our previous works, we assessed the reproducibility of cardiac magnetic resonance feature tracking (CMR-FT) strain parameters in a Landrace pig cohort during different inotropic states and one of them was due to dobutamine infusion using the same threshold (25% increase in HR) [16].

MR imaging was performed on a 3T clinical MR scanner (Ingenia, Philips Healthcare). 4D PC data were acquired with a 3D T1-weighted fast field echo (FFE) sequence with flow encoding gradients in three orthogonal axes (FH, RL, AP), in combination with retrospective gating to the electrocardiograph (ECG) cycle with 25 heart phases. Data were acquired in sagittal orientation, covering the entire heart and outflow tract. An anterior- and posterior phased array coil was employed for signal reception which consists of a flexible anterior and a posterior part that is integrated into the patient bed. Up to 28 coil elements were used for signal reception, depending on the size and position of the visual field recorded.

Typical scan parameters were as follows: Acquired FOV FH/RL/AP = 180 × 87 × 288 mm³,acquired resolution = 2.8 × 2.8 × 2.8 mm³,reconstructed resolution = 1.5 × 1.5 × 2.8 mm³, velocity encoding (VENC) along all three axes = 250 cm/s, TR/TE/flip = 3.8 ms/2.4 ms/5°, SENSE acceleration factor 2, bandwidth 2500 Hz / pixel.

4D flow data were acquired during normal mechanical ventilation. Respiratory gating or motion correction were considered unnecessary as respiration-induced bulk cardiac motion was found to be minimal in the animal cohort. The scan time was on the order of 10 min. After the MRI measurements were concluded, the animals were transported back to the operating room for sacrifice.

The resulting magnitude image and three velocities encoded images were imported in the software MEVISFlow (Fraunhofer MEVIS, Bremen, Germany) [26]. Pre-processing was applied, with noise-masking, antialiasing, automatic correction for eddy currents, and phase unwrapping as provided by MEVISFlow [27].

It became apparent that not all imported scans started at the correct time in the cardiac cycle due to incorrect triggering on the magneto-hemodynamic (MHD) effect, particularly under stress condition. This was manually corrected by rearranging the timepoints. Afterward, the pulmonary trunk (PT) and pulmonary arteries were located and segmented semi-automatically using an interactive watershed transform on the PC-MRA, which resulted in a 3D mask. Regions of interest (ROI) were placed by manually encircling the vessel of interest at three locations (Fig. 1): in the PT, before the vessel starts to dilate, at the left pulmonary artery (LPA) just after the first branch and in the right pulmonary artery (RPA) at the same height as in the LPA. Multiplanar reformatted images (MPRI) at the same location in the TP, LPA and RPA are defined manually for rest and stress. ROIs were defined manually and were automatically transferred motion compensated to all timeframes. The propagated ROIs were manually checked for each timeframe and corrected if necessary.

Placement of ROIs in pulmonary arteries in our pig’s cohort (n = 9). Represented in white is the main pulmonary trunk (PT), in blue the left pulmonary artery (LPA), while in red the right pulmonary artery (RPA). A proximal branch of the LPA has been detected in 7 out of 9 animals, ROI represented in yellow

The hemodynamic parameters obtained were:

NFD is defined as the distance between the center of the flow (\(\overrightarrow_}\)) and the center of the vessel (\(\overrightarrow_}\)), normalized by the vessel diameter, following the definition by Sigovan et al. [28]. A value of 0 means the flow is centered, and 1 means the flow is not centered. NFD is described by the following formula:

\(_=\frac_}- \overrightarrow_}\right|}_}\) where

$$\overrightarrow_}= \frac^_}\overrightarrow_} \times \left|\overrightarrow_}\right|}^_}\left|\overrightarrow_}\right|} \text\text\text \overrightarrow_}=\frac_^_}\overrightarrow_}}_}$$

TFD is defined as the average ratio of through-plane velocity magnitudes (\(}_\)) and the sums of through-plane \(\left(}_\right)\) and in-plane (\(}_\)) velocity magnitude. TFD is a measure of the flow swirl, which, unlike similar parameters like helicity, incorporates vessel orientation [28]. TFD is described by the following formula:

\(_= \frac_}\sum _^_}\frac}_\right|}}_\right|}^+}_\right|}^}}\) where

$$}_=\left(\overrightarrow_} \times \overrightarrow_}\right) \overrightarrow_} \text\text\text }_=\overrightarrow_}-}_$$

For evaluation of laminarity, minor flows superimposed to the main predicted flow pattern are evaluated and a value of 0 means there is no in-plane motion (no turbulence).

The angle measured in degrees was defined as the flow angulation in relation to the plane defined by the ROI area.

Data were analysed using Microsoft Excel and IBM SPSS Statistics version 23.0 software (SPSS Inc., Chicago, IL, USA) for Windows. Figures were made with GraphPad Prism version 8. All data are presented as mean ± standard deviation (SD). The Shapiro–Wilk test was used to determine whether the data were normally distributed. Data between groups at different inotropic states were analysed by one-way ANOVA for repeated measurements. Post-hoc testing was performed by Tukey’s test. Nonparametric variables were compared using the Wilcoxon test. A p-value of < 0.05 was considered statistically significant.

Inter- and intra-observer reproducibility was quantified using intra-class correlation coefficient (ICC). Agreement was considered excellent for ICC > 0.74, good for ICC 0.60–0.74, fair for ICC 0.40–0.59, and poor for ICC < 0.40. Data analysis was repeated after four weeks to assess intra-observer agreement. All the operators took the measurements twice, and the average values were taken. The agreement between the measurements was further assessed with Bland-Altman analysis, investigating both intra- and inter-observer agreements [29].