Neonatal Cardiac MRI
Despite many improvements in care, infants born extremely prematurely remain at high risk of death and neurodevelopmental impairment. The final goal of this project is to reduce mortality and morbidity through improving support of the cardiovascular system. We hope to achieve this by, for the first time, developing cardiovascular magnetic resonance (MR) techniques to study the preterm heart.
While the overall care provided to sick newborn infants has improved dramatically in recent decades, the most prematurely born infants remain at high risk of death and neurodevelopmental impairment. There is considerable evidence that circulatory factors play a central role in the pathophysiology of patterns of brain injury that result in poor long term outcome in preterm infants. However despite the importance of circulatory factors in determining outcome following premature birth, currently available clinical methods for assessing circulatory adequacy in preterm infants are inadequate due to the complex nature of the unique circulation present during the transition from fetal to extrauterine life. Blood pressure remains the most commonly monitored indicator in neonatal units, despite its uncertain correlation with volume of blood flow. Other clinical assessments such as capillary refill time, volume of urine output, etc have limited value in indicating circulatory health.
While a number of research tools for assessing circulatory status in preterm infants have been investigated, none have so far proven reliable in assessing cardiac function or cardiac output. While important advances in understanding of physiology have been made using echocardiography, the technique has poor repeatability, and is unable to accurately determine how well the heart muscle is actually contracting.
Cardiac magnetic resonance techniques have significantly advanced understanding of cardiovascular physiology and pathophysiology in adults. These non-invasive assessments of cardiac health are now being gained faster, in more detail and with greater sophistication than ever before. Techniques currently available in adults can accurately and reliably assess both contractility of the heart muscle and volume of blood ejected by the heart. In addition, tagging of distinct areas of heart tissue with a magnetic label has allowed analysis of the complex contractile and rotational movements seen within the heart. However until now technical challenges have prevented the use of cardiac MR in vulnerable preterm infants.
Advances in understanding of the physiology and pathophysiology of the heart and circulation in preterm infants are urgently required, as are improvements in monitoring and support of the failing heart. Cardiac MR techniques developed specifically to study the unique circulation of the preterm infant could provide real advances in knowledge, and ultimately lead to improvements in clinical outcome.
Performing Cardiac MR Imaging in the Newborn
We have now performed cardiac MR examinations in over 100 newborn preterm and term infants without any adverse events. Scans are carried out using a dedicated neonatal scanner installed within the Neonatal Intensive Care Unit at Hammersmith Hospital (Philips 3.0 Tesla Achieva system (Best, Netherlands)). Our group's process of safely performing MR imaging in preterm newborns has recently been described (Merchant N, Groves A, Larkman DJ, et al. A patient care system for early 3.0 Tesla magnetic resonance imaging of very low birth weight infants. Early Hum Dev. 85:779-783)
In all cases cardiac MR images are obtained after infants have been allowed to fall into a natural sleep after a feed, with careful swaddling. Sedative medication or anaesthesia have not been required. Infants are scanned with oxygen saturation, heart rate and continuous temperature monitoring, with a paediatrician or trained neonatal nurse in attendance throughout each scan. Scans can be performed free-breathing, with the provision of nasal continuous positive airway pressure (NCPAP) or low flow oxygen as clinically indicated. Protection from acoustic noise is achieved by applying moldable dental putty to the ears and covering them with neonatal ear muffs (Natus minimuffs, Natus Medical Inc., San Carlos, CA).
High quality images of cardiac function are obtained in infants weighing as little as 600 grams. To give an idea of the unique challenges in obtaining satisfactory images from the tiny preterm neonate, scale images from an adult and a neonate weighing just 590 grams are shown in figure 1.
Figure 1 – Scale comparison of cardiac four chamber views in a 70kg adult and 590 gram preterm infant.
In adult and paediatric cardiac magnetic resonance imaging most imaging techniques require image acquisition to occur either during episodes of breath-holding, or by use of a navigator bar to coordinate image acquistion with diaphragmatic movement. Both such techniques have significant impact on time required for image acquisition. A somewhat surprising advantage of neonatal cardiac imaging has been that high quality images have been obtained without the need for such measures. This improves the applicability of scanning in the neonatal population (by avoiding the need for intubation and ventilation) and reduces image acquisition time.
Most information on cardiac function can be obtained using very rapid balanced fast field echo (bFFE) sequences which have a temporal resolution of as little as 10 milliseconds. The standard imaging parameters used for bFFE sequences are shown in the table.
|Voxel Size (recon)||0.87 mm|
|Slice thickness||3-5 mm|
|Slice gap||0.3-0.5 mm|
|Phases per cycle||32|
Views obtained with bFFE sequences
Four chamber view – demonstrating left and right atria and ventricles, similar to echocardiographic apical four chamber view
Figure 2 – bFFE four chamber (A) and short axis (B) views in healthy newborn infant
(Figure 3, Movie file - Four chamber bffe view in newborn infant , 2.1MB, opens in new window). View used predominantly to allow tracking of atrioventricular valves through the cardiac cycle to eliminate inaccuracy from the through-plane motion of the base of the heart. This base-to-apex shortening may also be quantified using myocardial tagging techniques (see below).
Short axis view – demonstrating left and right ventricles, similar to echocardiographic parasternal short axis view. (Figure 4, Movie file - Short axis bffe view in newborn infant , 2.6MB, opens in new window). The critical view when examining cardiac function. Views taken at a single level allow analysis of wall thickness and thickening through the cardiac cycle. The position of this single slice can be determined from the previously acquired four chamber view, allowing for constant positioning at the mid-ventricular level. This permits enhanced accuracy of CMR over traditionally echocardiographic assessments, where small variaion in transducer positioning can lead to significant variation in quantitative measures in the tiny preterm heart.
However a ‘stack’ of adjacent slices may also be acquired to allow assessment of global cardiac function. Short axis stacks therefore allow assessment of total left and right ventricular volumes at end-diastole and end-systole by tracing the endocardial border in each slice. Cardiac MR therefore provides enhanced accuracy over echocardiography as the 2 dimensional nature of echocardiography means that assumptions, which are often inaccurate, have to be made about the shape of the ventricular cavities.
Specialist cardiac MR software is available to analyse ventricular function from stacks containing the entire cardiac volume (CMR Tools, Cardiovascular Imaging Solutions Limited, London). This software allows modeling of a mesh of the endocardial or epicardial borders (Figure 5, Movie file - Left ventricular model tracing endocardial and epicardial motion in healthy 1.5 kg infant , 3.3MB, opens in new window), and as mentioned above allows simultaneous tracking of the level of the atrioventricular valves eliminating inaccuracy from through-plane motion of the base of the heart
Figure 6 – CMR Tools allows tracking of the plane of the A-V valve.
Using bffe sequences and CMR Tools software we have been able to demonstrate that CMR shows approximately twice the repeatability of quantitative echo measures for measuring left ventricular output. In addition it allows assessment of volume of cardiac filling (preload) and ejection fraction (contractility) which have not previously been readily assessed by echocardiography.
In addition we have begun to describe the normal ranges of left and right ventricular output in stable preterm and term infants. Figure 7 demonstrates that absolute left ventricular output is closely associated with wieght at scan (A) and corrected gestational age (B). Weight corrected LVO (ml.kg-1.min-1) is lower at higher corrected gestational age (C), and range of values for LVO (ml.kg-1.min-1) is wider at lower postnatal age (D). The open circles in figures C and D represent infants with patent ductus arteriosus who have LVO well outside the normal range for stable infants. Similar ranges have also been described for right ventricular output, as well as ventricular filling and ejection fraction.
Figure 7 - Left ventricular output in 37 stable newborn infants (closed circles) and 3 infants with patent ductus arteriosus (open circles). Figures A and B show absolute LVO (ml.min-1), figures C and D show weight corrected LVO (ml.kg-1.min-1).
2-dimensional quantification of flow with phase contrast
Phase contrast techniques allow quantification of volume of flow in any large blood vessel. In contrast to echocardiography where the imaging plane is limited by ultrasonic window, a phase contrast slice can be placed in any orientation. Phase contrast sequences can be externally validated against flow phantoms to confirm accuracy of quantitative measures, and also allow internal validation of cardiac output (by comparing measurements taken by bFFE and phase contrast techniques). However their particular value in the neonate lies in quantifying flow at multiple points in the circulation. The persistence of fetal shunt pathways in the preterm neonate means that neither left nor right ventricular output can necessarily be taken to represent true systemic or pulmonary perfusion. Cardiac MR allows quantification of flow in the superior vena cava (SVC) and descending aorta (DAo), both of which are considered to be markers of true systemic perfusion in the preterm neonate. In the image shown below, static tissues are shown as mid-grey, flow in the head-foot direction is dark, flow in the foot-head direction is light.
Figure 8 – Phase contrast image in axial plane showing SVC and DAo.
(Figure 9, Movie file - Phase contrast movie showing quantifiable SVC and DAo flow , 1.5MB, opens in new window). By estimating velocity in each pixel covering the vessel throughout the cardiac cycle a velocity-time graph is produced, with the area under the curve representing total volume of flow:
Figure 10 – Quantification of SVC and DAo flow from phase contrast sequences
We have demonstrated that 2-d phase contrast quantification of flow in the newborn aorta correlates well with bffe measures of left ventricular output; that quantification of SVC and DAo flow shows approximately twice the repeatability to that seen in prior echocardiographic cohorts; and that when the ductus arteriosus has closed the sum of SVC and DAo correlates closely with left ventricular output, suggesting that SVC + DAo flow is a reasonable surrogate for total systemic perfusion.
3-dimensional visualisaion of flow with phase contrast
While 2-dimensional phase contrast techniques allow quantification of flow wihin a single blood vessel, 3-dimensional techniques allow visualisation of flow in entire regions of the body. Specialist post-processing programmes have been developed that allow the user to trace the path of a notional bolus of blood from within any vessel, throughout the cardiac cycle (so-called 4-dimensional flow imaging - 3 spatial dimensions plus time). We have applied 3-dimensional phase contrast sequences in newborn preterm and term infants, and utilised Gyrotools software (GyroTools Ltd., Switzerland) to visualise flow in the aortic arch (Figure 10, Movie File - Term infant, aortic arch 0.7MB, opens in new window) and pulmonary arteries (Figure 11, Movie File - Term infant pulmonary bifurcation , 0.7MB, opens in new window).
Figure 12 - 3-dimensional phase contrast visualistion of blood flow in the aortic arch of a newborn infant.
In addition, 3-dimensional imaging techniques provide scope for answering more complex questions about haemodynamics in the newborn. Kilner et al (Nature 2000;404:759-761) initially demonstrated that in the healthy adult, blood flowing into the right atrium displays a characteristic rotational motion which is felt to conserve the kinetic energy of blood. Using 3-dimensional phase contrast images we have been able to confirm the presence of these flow patterns in healthy adults (Figure 13, Movie File - Healthy adult rotational RA flow , 0.5MB, opens in new window).
We have also been able to demonstrate that some newborn infants show a degree of anti-clockwise rotational flow similar to the adult pattern (Fig 14, Movie File - 1kg infant, rotational RA flow , 0.7MB, opens in new window).
Figure 15 - 3-dimensional phase contrast visualistion of rotational right atrial blood flow in an adult (left) and a preterm infant (right).
However in some newborn infants there is no evidence of rotational RA flow, despite flow being consistently tracked through to ventricular filling (Figure 16, Movie File - Term infant, chaotic RA filling , 1.6MB, opens in new window). This altered pattern from the adult model could be produced by protrusion of the Eustachian valve from the posterior wall of the right atrium, or by shunting through the patent foramen ovale. The disrupted patterns of flow seen in some newborns may mean that these infants do not benefit from the momentum-preserving effects of the characteristic rotational flow patterns seen in adults, potentially leading to impaired ventricular filling and cardiac function in the perinatal period.
Myocardial tagging techniques
Myocardial tagging techniques use pulses of radiofrequency to transiently saturate myocardial tissue along set lines or grids, producing low signal areas. These ‘tags’ are then distorted by the motion of the cardiac musculature. Our initial attempts at tagging have utilised steady state free procession (SSFP) image acquisition techniques.
In its most basic form tagging allows quantification of the base-to-apex shortening of the heart (Figure 17, Movie file – Linear tagging sequence on four chamber view tracking base-to-apex shortening , 0.6MB, opens in new window). Measurement of base-to-apex shortening is of particular relevance as impaired axial contractility is considered to be an early indication of global dysfunction as the longitudinal fibres that produce this shortening lie predominantly in the endocardial layer which has the most fragile blood supply.
It is now appreciated that as well as shortening in the radial and axial directions, the myocardial tissue undergoes a complex rotational motion with each heart beat, ‘wringing’ blood out from the chambers. Our early studies suggest that this process is present, at least in healthy preterm infants, such that when visualised in the short axis view the base of the heart rotates clockwise during systole (Figure 18, Movie file – Grid tagging sequence at base of heart on short axis view showing clockwise systolic motion , 0.6MB, opens in new window), the apex of the heart rotates anticlockwise during systole.
However SSFP tagging sequences have significant limitations - tag persistence is poor, as is temporal resolution, and there is as yet no widely available automated analysis techniqe, leaving the process vulnerable to errors from poor repeatability of analysis. Workers at Imperial College have demonstrated that it may be possible to produce semi-automated analysis of SSFP tagged images of myocardial motion in adults using image tracking techniques (Figure 19, Movie file - Automated tag tracking of healthy adult on short axis view , 9MB, opens in new window). Images courtesy of Hui Xue, Imaging Sciences Department, Imperial College London. While resolution of SSFP tagged images is currently poorer in neonates, preliminary work suggests that similar techniques may also be viable in this population (Figure 20, Movie file - Automated tag tracking in neonate on short axis view , 4MB, opens in new window). Images courtesy of Hui Xue, Imaging Sciences Department, Imperial College London.
An alternative approach may be to employ different image acquisition sequences. Complementary spatial modulation of magnetisation (CSPAMM) has been shown to have improved tag persistence and temporal resolution in adults, and has the additional advantage that automated analysis with harmonic phase (HARP) techniques are widely available. We have successfully applied CSPAMM images sequences to adult volunteers at 3 Tesla (Figure 21, Movie file - Adult CSPAMM Image at 3 Tesla , 6MB, opens in new window) and are in the procees of adapting image acquisition for use in neonates.
Strengths of MR Assessment of Cardiac Function
While some advances in understanding of the pathophysiology of the neonatal circulation have been made using studies with echocardiography, the technique is far from ideal. In particular the relatively poor repeatability of echo measures of flow limits its use as an end-point in clinical trials of intervention. Cardiac MR techniques have a number of advantages over echocardiography:
•Cardiac MR is less operator dependent, enhancing repeatability
•Direct visualisation of function in all areas of the heart obviates the need for assumptions of cardiac geometry which are often overly simplistic
•Assessment of shortening in radial and axial planes, along with assessment of rotational motion provides multiple markers of contractility. Placement of these image slices can determined from orthogonal views, reducing error in quantitative measures from variability in image plane. This may be of particular value in assesment of velocity of circumferential fibre shortening, an afterload-correctable measure of 'true' contractility taken from the short axis view. Afteroad correction of functional measures is of particular significance in neonates given the prominent role of peripheral vascular resistance in governing blood pressure in this population.
•Multiple techniques allow internal validation of measurements of flow
•Improved repeatability allows decreased subject numbers while maintaining the power of interventional studies
•Multiple assessments of intracardiac volumes, wall motion and blood flows will provide an enhanced appreciation of how clinical interventions impact on the three principle components of cardiac function – preload, contractility and afterload
The final goal of this project is to reduce mortality and morbidity in the preterm infant through improving support of the cardiovascular system. We aim to achieve this goal by firstly developing a comprehensive, systematic approach to the assessment of cardiac function in newborn infants using MR. It has previously not been possible to measure intracardiac flow or obtain precise data on myocardial function in preterm infants, still less to develop an integrated understanding of the inter-related effects of cardiac filling, contractility and afterload.
Such an approach will allow increased understanding of cardiovascular physiology and pathophysiology in the unique circulation of the preterm neonate. Most significantly we feel that the complex global information gained about the circulation by a systematic MR assessment of function will allow MR to act as a comprehensive end-point in clinical trials of circulatory intervention. The power of the MR assessment should enable our group to detect significant differences in circulatory outcome by studying relatively small numbers of infants.
In addition, by studying how clinical and echocardiographic markers of cardiac function relate to reliable MR assessments we hope to enhance the ability of all clinicians to reliably assess the circulation at the cot-side. This will allow the results of the studies described above to be generalised to the majority of neonatal units where MR assessment of cardiac function is unlikely to be routinely available for some time.
In particular advanced echocardiographic techniques such as tissue Doppler and speckle tracking modalities have promise to provide significant improvements in cotside evaluation by reliably assessing 'true' contractility. However the techniques have very limited validation in the neonatal population. In our future work we will perform standard and advanced echocardiographic study alongside CMR assesments. Comparing tissue Doppler imaging at multiple sites with CMR measures of function both in population studies and during trials of circulatory intervention will allow identification of optimal correlation with CMR measures and thorough assessment of repeatability of techniques.
We are confident that current weaknesses in circulatory monitoring and support contribute to the high risk of adverse outcome in extremely preterm infants, and that by improve circulatory monitoring and support our group will enable significant improvements in survival and neurodevelopmental outcome in vulnerable preterm infants.
David Edwards - Professor of Neonatal Medicine, Faculty Theme Leader, Imaging Science
Dr Alan Groves - Clinical Senior Lecturer in Neonatal Medicine
Ms Giuliana Durighel - Research Radiographer
Professor JV Hajnal - Chair in Imaging Science, Imperial College London
Professor Reza Razavi - Division of Imaging Sciences, King's College London