Feasibility of Fetal Cardiac Magnetic Resonance Imaging: Preliminary Experience

The role of fetal magnetic resonance imaging (MRI) as a tool additional to ultrasound in the complete work-up of the fetoplacental anatomy has grown exponentially since the first reports in 19851-6. Unlike ultrasound imaging, this modality is not affected by maternal and fetal conditions such as obesity7 and oligohydramnios8, which particularly impair sonographic visualization of the fetal heart; maternal obesity increases the rate of sub-optimal ultrasound visualization of the fetal cardiac structures by 49.8%7, despite advanced ultrasound equipment9. visceroatrial situs, assessed in relation to the bronchial tree: the embryologically normal left main bronchus is long with no early division and runs under the left pulmonary artery, whereas the right one is short, more vertical, has an early division and runs behind the right pulmonary artery; ventricular looping: in a normal D-loop, the most anterior ventricle is the embryological right one, with a trabeculated septal wall, and the most posterior is the embryological left one, with a smooth septal wall; ventriculoarterial connections, i.e. the normal crossing of the origin of the great vessels, with the main pulmonary artery crossing above the aortic root in a sagittal plane, and the aortic arch mostly parallel to the left pulmonary artery. a balanced four-chamber view; the side of the aortic arch, defined according to the main bronchus above which it crosses the mediastinum anteroposteriorly; the systemic and pulmonary venous connections: both superior and inferior vena cava should be seen joining the right atrium, and at least one pulmonary vein should be seen joining the left atrium. For both MRI and ultrasound, each anatomical component was classified as 'normal', 'abnormal' or 'not visible'. Institutional review board approval was obtained to access the postnatal data of all patients. The mean gestational age at MRI was 29.3 (range, 24–34) weeks. Ultrasound was performed on average 6.4 (range, 4–9) days before MRI. No patient had a body mass index consistent with obesity, and there were no cases of oligohydramnios. On analysis of ultrasound data, all cardiac anatomical components were classified as normal in all cases, but on analysis of MRI data, all were classified as normal in eight out of ten cases. In two patients all items were categorized as not visible, apart from the balanced four-chamber view, which was classified as normal. The RARE acquisition sequence was particularly useful in the assessment of situs (Figure 1). Ventricular looping was identified in all fetuses thanks to ventricular morphology on T2-weighted true-FISP sequences (Figure 2). Ventriculoarterial connections and the four-chamber view were best analyzed on a sequential analysis of axial T2-weighted true-FISP images (Figure 3). The side of the aortic arch could be visualized in the axial plane on T2-weighted true-FISP images, as well as on coronal T2-weighted HASTE images, mimicking the normal left para-aortic line (i.e. the normal left paravertebral course of the thoracic aorta) (Figure 4). T2-weighted true-FISP images also enabled the most accurate visualization of venous connections, in coronal or sagittal planes for the venae cavae (Figure 5a), and in the axial plane for the pulmonary veins (Figure 5b). T1-weighted images were not useful. Coronal RARE sequence showing the fetal visceroatrial situs, assessed in relation to the bronchial tree: the embryologically normal left main bronchus (with the stomach on the same side) is rather long with no early division (large arrow), whereas the right one is short, more vertical, and has an early division (small arrow). Axial true-FISP section demonstrating fetal cardiac ventricular morphology: the most anterior ventricle is the embryological right one, with a trabeculated septal wall (arrow), and the most posterior is the embryological left one, with a smooth septal wall. Sequential analysis of axial true-FISP sections (a–d) from the inferior to the superior parts of the fetal chest: there is normal crossing of the origin of the great vessels, with the main pulmonary artery (arrow) crossing above the aortic root (*) in a sagittal plane, with the aortic arch (arrowhead) mostly parallel to the left pulmonary artery. Coronal HASTE image illustrating determination of side of the aortic arch, defined according to the main bronchus above which it crosses the mediastinum anteroposteriorly, mimicking the normal left para-aortic line (arrowhead), as usually described on postnatal chest X-ray. (a) Systemic venous connections (arrowheads) as seen on the coronal plane in T2-weighted true-FISP image. (b) One or two pulmonary veins (arrowheads) were visible on axial true-FISP images. No congenital heart defect was diagnosed by sonography or MRI. All infants were alive and well at the time of writing, with a normal physical cardiac examination. When MR images were suitable for analysis, concordance with ultrasound was 100%. Fetal movements in the two earliest cases were the main reason for the lack of visualization. Both fetal and cardiac MRI have improved dramatically, thanks to the availability of fast sequences and cardiac gating, making MRI one of the most powerful approaches to imaging congenital heart diseases12. A study comparing MRI and three-dimensional ultrasound examination of fetal cardiac specimens, showed MRI to have the better image quality, with more structural details13. Artifacts due to fetal movements are still a concern in MRI, but in most situations the very short acquisition time can be used in association with the mother's cooperation, through repeated short (10-s) apneas, without compromising the image quality. Maternal sedation is rarely required. From a technical perspective, further improvement in fetal heart MRI will require synchronization to the fetal heart rate. This was described recently in two cases14, but is not yet available in clinical practice. There are very few reports in the literature on fetal MRI of cardiac structures15, 16. The most convincing concerns a cardiac rhabdomyoma15, but MRI was limited to single-shot imaging and had limited spatial resolution. Here we have provided preliminary evidence that fetal cardiac examination is feasible using 'routine' fetal MRI protocols, without special techniques such as fetal cardiac gating, and that the most efficient sequence is T2-weighted true-FISP imaging. This could be a helpful tool in addition to ultrasound in situations of maternal obesity or oligohydramnios. Fetal cardiac gating devices may soon be available in clinical practice, or there is the possibility of maternal sedation in selected cases of fetal cardiac MRI. What the future has in store may depend on recent developments of this modality, such as real-time 'cine' fetal MRI (Videoclip S1, online). Many thanks to Dr Chantale Lapierre (Sainte Justine Hospital, Montreal) who taught me everything I know about the heart, and to the MRI technologists of our department for their thoughtful help. The following material is available from the Journal homepage: http://www.interscience.wiley.com/jpages/0960-7692/suppmat (restricted access) Videoclip S1 Real-time magnetic resonance imaging (MRI) across the fetal chest at the level of the cardiac ventricles. The sequence is based on balanced-fast-field-echo imaging, imported from our pediatric cardiac MRI protocol, with no fetal heart-beat gating. Both maternal and fetal breathing movements can be observed. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.