Cranial ultrasound on neonatal intensive care units is generally performed by intensive care physicians, but radiologists often provide this crucial bedside test to children on specialist paediatric cardiac intensive care units. On a paediatric cardiac intensive care unit, complex congenital cardiac conditions are commonly encountered in both pre- and postoperative scenarios, often with the use of extracorporeal membrane oxygenation (ECMO), which both increases the risks of a number of neurologic complications and results in significant changes in vascular physiology. The aim of this pictorial essay is to discuss cranial ultrasound technique, demonstrate the changes in Doppler flow profiles resulting from veno-arterial extracorporeal membrane oxygenation and congenital cardiac conditions, and illustrate commonly encountered intracranial complications of extracorporeal membrane oxygenation support in congenital cardiac care.
In the United Kingdom, most cranial ultrasound (US) examinations are performed in the context of prematurity, by neonatal intensive care physicians. However, in specialist paediatric cardiac surgical centres, radiologists often provide this crucial bedside imaging service. In these units, veno-arterial extracorporeal membrane oxygenation (ECMO) is commonly used as a form of periprocedural support in neonates with particularly complex congenital heart diseases.
These children sometimes sustain significant neurologic complications, both from their underlying cardiac disease and subsequent surgical intervention, as well as from anticoagulation and the ECMO support itself. A two-decade review of neurologic events related to ECMO in The Netherlands reported cranial ultrasound abnormalities in 17% of patients, most frequently haemorrhage (8.8%) and ischaemic stroke (5%), with higher frequencies of complications in veno-arterial ECMO and an interesting predominance of left-sided pathology .
Extracorporeal membrane oxygenation support is not compatible with MRI, and at times safe transfer to the CT scanner is not possible. Outside of the few centres employing mobile CT units, this leaves ultrasound as the only immediately available modality. The aim of this pictorial review was to illustrate the US appearance of common and less common intracranial pathologies encountered in neonates and infants with congenital heart diseases, before, during and after receiving ECMO support.
Extracorporeal membrane oxygenation
In veno-arterial ECMO, blood is diverted from the child’s systemic venous system through an external membrane for gaseous exchange, and then the oxygenated blood is returned to the child’s arterial system at systemic arterial pressure, thus facilitating both oxygenation and systemic vascular output (Fig. 1). The catheters are either placed peripherally (usually via the right common carotid artery) or centrally through a median sternotomy created during surgery (Fig. 2). The former approach requires ligation of the right common carotid artery, which can be later reconstructed after decannulation.
Anticoagulation is required to prevent the formation of thrombi within the ECMO circuit. Together with altered haemodynamics and damage to, or occlusion of, the neck vessels, there is an increased risk of neurologic complications, including intracerebral haemorrhage and ischaemic/embolic infarction .
Relevance of cranial ultrasound in extracorporeal membrane oxygenation support
Cranial US is a safe, portable technique capable of real-time structural assessment and colour phase and spectral Doppler imaging of intracranial arterial and venous flow. It is often used to provide baseline imaging before cardiac surgery, as a bedside monitoring tool after initiation of ECMO support, and for daily follow-up of evolving pathology. If further characterisation of lesions detected on US is required or clinical suspicion persists despite a negative US, CT is the cross-sectional imaging modality of choice because the ECMO equipment is not compatible with the MRI scanner. After ECMO therapy is withdrawn, MRI is the follow-up imaging modality of choice. There are, of course, limitations to cranial US, with the diagnostic quality of cranial US being highly operator-dependent and with reduced visualisation of the posterior fossa and convexity regions. Furthermore, ultrasound has reduced sensitivity and specificity for pathologies other than major haemorrhage compared to both CT and MRI [3,4,5,6].
The anterior fontanelle provides the largest cranial sonographic window for both coronal and sagittal image acquisition and is usually the last of the fontanelles to close (50% close by 16 months of age) . In addition to grey-scale anatomical images, colour and spectral Doppler assessment can be obtained of the intracranial internal carotid and basilar arteries and the circle of Willis, as well as the superior sagittal sinus, vein of Galen and straight sinus (Fig. 3). Supplemental views of the posterior fossa can be obtained via the posterior and mastoid fontanelles  (Fig. 4).
In a normal brain, low-resistance waveforms should be evident within both intracranial internal carotid arteries with vertical systolic upstrokes (short acceleration time) and constant forward flow throughout diastole (Fig. 5). In a child on ECMO, ligation of the right common carotid artery results in absent or reversed flow in the right intracranial carotid artery and non-pulsatile forward flow in the other intracranial arteries from the continuous non-pulsatile output of the ECMO pump (Fig. 6). With the right common carotid artery occluded, supply to the right hemisphere occurs via the circle of Willis, with retrograde flow through the right posterior or anterior communicating arteries. This retrograde flow might persist after decannulation from extracorporeal membrane oxygenation support, even following surgical reconstruction of the common carotid artery at decannulation  (Fig. 6d and e). Reassuringly, even in cases of persistent internal carotid stenosis post reconstruction, the vast majority of children retain symmetrical cerebral perfusion, in keeping with good collateralisation and compensation via the circle of Willis [9, 10].
Intracranial haemorrhage is the most frequent neurologic complication in neonates receiving ECMO support , the majority of cases occurring within 72 h of initiation of ECMO support . It is notable that most non-survivors of ECMO support die from intracranial haemorrhage rather than their primary cardiorespiratory disease .
Intracranial sites of haemorrhage visible on ultrasound include the germinal matrix, ventricular system, cerebral and cerebellar parenchyma and extra-axial spaces (subarachnoid, subdural and extradural spaces). On ultrasound, haematoma normally appears as echogenic (bright) space-occupying material (Fig. 7), and depending on the volume and location of haemorrhage, it might be accompanied by midline shift or dilatation of the lateral ventricles. If patient transfer is safe, the extent of haemorrhage should ideally be confirmed via subsequent CT, which shows haemorrhage as hyperattenuating material (Fig. 8).
Intraventricular haemorrhage can obstruct arachnoid granulation tissue, resulting in communicating hydrocephalus (Fig. 9), at times requiring neurosurgical intervention. The degree of hydrocephalus can be monitored via serial ultrasonography with measurement of the ventricular index (e.g., Levene index) or dimensions/area of the lateral ventricular horns.
As haematoma begins to liquefy, particularly in the extra-axial spaces, it can become hypo- or even anechoic, often containing echogenic debris or even a debris–fluid level (Fig. 10). Serial ultrasound examinations can track the evolution of previously confirmed haematoma without the need for repeated CT or MRI examinations.
After haemorrhage, the second most common major neurologic injury sustained by infants on extracorporeal membrane oxygenation is ischaemia, whether secondary to underlying pathology (e.g., complex cyanotic congenital heart disease), complications of surgery or initiation of ECMO support (emboli, ligation of the common carotid artery with insufficient collateral flow) or a combination of factors. On ultrasound, regions of established ischaemic infarction appear hyperechoic, with loss of grey–white matter differentiation occasionally appreciable on high-frequency imaging. Infarcts often occupy characteristic vascular territories but are occasionally more localised (either as small wedge-shape infarcts or focal lacunar infarcts). If small and localised they can be difficult to distinguish from focal haemorrhage, in which case correlation with cross-sectional imaging (CT if continued extracorporeal membrane oxygenation support is required) is useful (Fig. 11).
Venous infarcts typically occur in a non-arterial distribution and are most frequently visible as hyperechoic regions in the thalami or basal ganglia (Fig. 12). Infarcts can undergo haemorrhagic transformation, with the echogenic appearance of both infarct and haemorrhage on ultrasound, leading to potential confusion. In cases of diagnostic uncertainty, further interrogation with CT is helpful.
Cortical laminar necrosis
Cerebral energy depletion, described in conditions including infarction and prolonged hypoxic–ischaemic encephalopathy, can lead to necrosis of cortical lamina. This can be associated with diffuse, focal or punctate cortical haemorrhage and can later calcify . The appearance on ultrasound is not well described, but is likely to vary according to the presence and extent of necrosis. Unenhanced CT might demonstrate high-attenuation cortex with further cortical hyperenhancement following contrast administration (Fig. 13). Unenhanced MRI demonstrates high cortical T1 signal. Recent studies have shown that the high attenuation on CT and corresponding high T1 signal on MRI only rarely represent haemorrhage, as once thought, and likely relate to protein degradation within necrotic grey matter .
Periventricular leukomalacia is a white matter injury related to watershed ischemia of the immature cerebrovascular system, frequently seen in premature neonates. There are two different forms: diffuse and cystic. The diffuse form can be more difficult to detect on ultrasound and might require evaluation with MRI . The cystic form initially appears as increased periventricular white matter echogenicity, and later develops into multiple anechoic periventricular cysts (Fig. 14). It is important from a prognostication point because cystic periventricular leukomalacia is associated with a worse neurologic outcome .
Lenticulostriate vasculopathy is thought to be caused by thickening of the lenticulostriate branches of the middle cerebral arteries and leads to branching echogenic structures becoming visible within the basal ganglia on cranial US. The specific pathogenesis is not fully understood , although it is associated with a number of underlying disorders including congenital heart disease and hypoxic–ischaemic events . Unsurprisingly, it is therefore frequently seen in infants on ECMO support (Fig. 15).
Diffuse hypoxic–ischaemic encephalopathy
Severe diffuse hypoxic–ischaemic encephalopathy in neonates initially results in a central pattern of damage, involving deep grey matter (basal ganglia and thalami), but with prolonged ischaemia, the cortex can also become involved. Ultrasound might initially appear normal or demonstrate a mild global increase in cerebral echogenicity with or without diffuse swelling obliterating the cerebrospinal fluid spaces. Increased echogenicity of the thalami (Fig. 16) suggests more severe injury and is associated with a poorer outcome . Doppler examination of the anterior and middle cerebral arteries has been employed in the setting of suspected hypoxic–ischaemic encephalopathy, with a resistive index of less than 0.6 associated with poor outcome, even if other sonographic signs are absent .
Cranial US is a safe, portable technique allowing screening prior to, and monitoring whilst on, extracorporeal membrane oxygenation support. Whilst ultrasound is the first-line imaging technique in these children, significant limitations (particularly in the setting of suspected acute ischaemia) leave CT as the modality of choice for confirming and demonstrating the extent of significant pathologies whilst a child is on ECMO support, particularly when important treatment decisions are required, often including the decision as to whether to continue ECMO support.
Raets MMA, Dudink J, Ijsselstijn H et al (2013) Brain injury associated with neonatal extracorporeal membrane oxygenation in the Netherlands: a nationwide evaluation spanning two decades. Pediatr Crit Care Med 14:884–892
de Mol AC, Liem KD, van Heijst AFJ (2013) Cerebral aspects of neonatal extracorporeal membrane oxygenation: a review. Neonatology 104:95–103
van Wezel-Meijler G, Steggerda SJ, Leijser LM (2010) Cranial ultrasonography in neonates: role and limitations. Semin Perinatol 34:28–38
Rollins MD, Yoder BA, Moore KR et al (2012) Utility of neuroradiographic imaging in predicting outcomes after neonatal extracorporeal membrane oxygenation. J Pediatr Surg 47:76–80
LaRovere KL, Vonberg FW, Prabhu SP et al (2017) Patterns of head computed tomography abnormalities during pediatric extracorporeal membrane oxygenation and association with outcomes. Pediatr Neurol 73:64–70
van Heijst AFJ, de Mol AC, Ijsselstijn H (2014) ECMO in neonates: neuroimaging findings and outcome. Semin Perinatol 38:104–113
Pindrik J, Ye X, Ji BG et al (2014) Anterior Fontanelle closure and size in full-term children based on head computed tomography. Clin Pediatr 53:1149–1157
Di Salvo DN (2001) A new view of the neonatal brain: clinical utility of supplemental neurologic US imaging windows. Radiographics 21:943–955
Mitchell DG, Merton D, Desai H et al (1988) Neonatal brain: color Doppler imaging. Part II. Altered flow patterns from extracorporeal membrane oxygenation. Radiology 167:307–310
Wien MA, Whitehead MT, Bulas D et al (2017) Patterns of brain injury in newborns treated with extracorporeal membrane oxygenation. AJNR Am J Neuroradiol 38:820–826
Lago P, Rebsamen S, Clancy RR et al (1995) MRI, MRA, and neurodevelopmental outcome following neonatal ECMO. Pediatr Neurol 12:294–304
Niwa T, Aida N, Shishikura A et al (2008) Susceptibility-weighted imaging findings of cortical laminar necrosis in pediatric patients. AJNR Am J Neuroradiol 29:1795–1798
Murgo S, Avni EF, David P et al (1999) Periventricular leukomalacia in premature infants: prognostic role of ultrasonography and MRI. J Radiol 80:715–720
Shin HJ, Kim M-J, Lee HS et al (2015) Imaging patterns of sonographic lenticulostriate vasculopathy and correlation with clinical and neurodevelopmental outcome. J Clin Ultrasound 43:367–374
Lowe LH, Bailey Z (2011) State-of-the-art cranial sonography: part 2, pitfalls and variants. AJR Am J Roentgenol 196:1034–1039
Stark JE, Seibert JJ (1994) Cerebral artery Doppler ultrasonography for prediction of outcome after perinatal asphyxia. J Ultrasound Med 13:595–600
Conflicts of interest
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Svrckova, P., Meshaka, R., Holtrup, M. et al. Imaging of cerebral complications of extracorporeal membrane oxygenation in infants with congenital heart disease — ultrasound with multimodality correlation. Pediatr Radiol (2020). https://doi.org/10.1007/s00247-019-04603-1
- Congenital heart disease
- Extracorporeal membrane oxygenation
- Vascular imaging