Cardiac CMR, especially as a postoperative follow up modality, has become the primary diagnostic imaging tool for CHD patients. Not only because it is the gold standard for the evaluation of left and right ventricular function, but also because it allows for reliable assessment of vascular structures independent of an acoustic window. While recent studies have mostly focused on increasing the temporal resolution of MRA in patients with CHD [9, 10], imaging issues related to motion artefacts (respiratory and cardiac motion) have yet not been fully addressed. Especially in children and diseased patients limited compliance may deteriorate image quality of FP-MRA.
The combination of ECG- and respiratory navigator-gated MRA using a blood-pool contrast agent at 1.5 T has shown to overcome these limitations . However, blood-pool contrast agents are of higher costs and even more importantly, they exclude the possibility of viability imaging. Studies have revealed the value of viability imaging in patients after surgery of CHD as the presence of scar or fibrous tissue could be related to adverse ventricular mechanics and a higher prevalence of non-sustained ventricular tachycardia [4, 5].
Thus, an ideal imaging protocol for the assessment of complex congenital heart disease should not only include functional imaging and MRA but also viability imaging. In this study, a comprehensive imaging protocol was established, which includes functional imaging, FP-MRA, as well as viability imaging. Adding viability imaging does require the injection of a double dose of contrast agent as well as a waiting period of approximately 12–15 minutes after contrast injection to allow for delayed enhancement imaging. The second injection and waiting period were used to acquire a high-resolution motion compensated MRA sequence (HR-MRA) at no cost of additional scan time or contrast agent.
In comparison to the FP-MRA, the HR-MRA sequence revealed a significantly better image quality and vessel sharpness (Figures 1, 2, 3). Most likely, these results could be attributed to the higher resolution as well as the improved motion compensation through ECG-gating and navigator respiratory compensation. Another noteworthy advantage of HR-MRA compared to FP-MRA is the acquisition of isotropic voxels allowing for multiplanar image reconstruction without any loss of image resolution (Figure 1, 2). The fact that HR-MRA revealed an abnormal offspring of the left coronary artery, missed by FP-MRA in one patient, highlights the value of isotropic high resolution and motion compensation (Figure 3). One inherent drawback of the motion compensated approach is the missing dynamic information. Thus, it can only be used as an add on sequence. However, implementing the HR-MRA into a standard CHD-CMR-protocol does not involve any additional scan time or the use of additional contrast agent injection.
Due to the slow injection rate that was used for the HR-MRA, the relative contrast between the examined vessels and surrounding tissue in all acquired HR-sequences was lower compared to the bolus technique (FP-MRA). Due to the most compact bolus geometry in the PA during the first-pass, the largest difference between relative contrasts was found in this vessel.
Intra- and interobserver agreement results revealed much closer 95% limits of agreement and higher correlation coefficients for the HR-MRA sequence in comparison to the FP-MRA sequence (Figure 4). This shows the importance of vessel sharpness rather than relative contrast for accurate assessment of vessel diameters.
The MRA sequence employed in this study was similar to the approach by Naehle et al. . They compared a first-pass MRA-protocol with a high-resolution motion compensated steady-state protocol following the injection of a blood-pool contrast agent (Vasovist) at 1.5 T . The use of a blood-pool agent in their study did allow them to deliberately choose an inversion time for optimal suppression of background tissues, whereas in our study a fixed inversion time had to be used due to the limited acquisition window after contrast agent injection. Also, the spatial resolution had to be sacrificed to allow for faster acquisition of the 3D HR-MRA dataset. Nevertheless, vessel sharpness of the motion compensated MRA vs. the first-pass-MRA was very similar in both studies, using the same measurement technique. Similar to Naehle et al., an additional clinical finding was also detected by motion compensated MRA which was missed by FP-MRA, underscoring the value of motion compensation. Most importantly, the HR-MRA approach in this study overcomes a major limitation of the previously published approach as it allows for viability imaging and thus fits perfectly into a standard CHD-CMR-protocol. In addition, contrary to blood-pool contrast agents, the contrast agent used in this study has recently been approved for use in children > 2 years in North America and Europe.
Previous studies have demonstrated the value of 3D whole-heart imaging in pediatric populations using balanced steady-state free precession (b-SSFP) sequences [12, 13]. However, one advantage of the proposed approach in comparison to b-SSFP is the high CNR, which is achieved through the use of an IR-pulse suppressing the background signal. In addition, while the non-contrast enhanced b-SSFP approach has proven to work at 1.5 T, its use may be limited by artefacts related to b0 and b1 field inhomogeneity at 3 Tesla.
Another recent study by Yang et al.  demonstrated the feasibility and diagnostic yield of a contrast-enhanced whole-heart coronary MRA at 3 T using an extracellular contrast agent (gadobenate dimeglumine) during slow injection. Interestingly, although the contrast agent in this study (gadobutrol) has an even shorter plasma half-life than gadobenate dimeglumine, the acquisition of a 3D motion compensated HR-MRA dataset, covering the entire thorax with an even higher resolution, was feasible.
Using a fixed TI of 200 ms, Yang et al. achieved an optimal suppression of background tissue. In contrast, the TI chosen in this study was significantly longer for several reasons. First, the contrast agent could not be completely injected as a slow infusion, because a bolus injection is required for FP-MRA, which in turn results in a broader variation of contrast agent concentration in the blood. Second, the albumin binding of gadobutrol is even less, yielding an even shorter imaging window. Finally, the heart rate of the patients in this study was not controlled resulting in a wide range of heart rates. Thus, we decided to choose a longer TI, which might not reveal optimal suppression of background tissue but yields a good overall signal to noise ratio for the majority of the patients. Moreover for imaging of the entire thoracic vasculature, suppression of background signal is less critical compared to coronary imaging.
Due to variations in patients’ total amount of contrast agent injected, variation in heart rate, navigator performance and the subsequent variation of acquisition duration, it is difficult to determine the optimum contrast agent injection protocol for the proposed FP/HR-MRA approach. The use of a bolus for FP-MRA and subsequent slow injection of contrast agent has shown to deliver a good CNR in this study. However, further research is warranted to determine, whether a further improvement of CNR can be achieved with a tailored protocol taking into account the aforementioned variables.