What this paper adds
- • Knowledge of fetal development as visualized with fetal MR imaging.
- • Knowledge of common central nervous system pathologies as seen with fetal MRI.
- • Knowledge of the role of fetal MR in clinical practise.
Fetal magnetic resonance imaging (MRI) has become established as part of clinical practice in many centres worldwide especially when visualization of the central nervous system pathology is required. In this review we summarize the recent literature and provide an overview of fetal development and the commonly encountered fetal pathologies visualized with MRI and illustrated with numerous MR images. We aim to convey the role of fetal MRI in clinical practice and its value as an additional investigation alongside ultrasound yet emphasize the need for caution when interpreting fetal MR images especially where experience is limited.
For many years, ultrasound has been the mainstay for prenatal diagnosis of fetal abnormality because, until relatively recently, no other robust and reliable non-invasive method for examination of the unborn fetus existed. In the current climate of increasing levels of maternal obesity, multiple pregnancies from assisted reproduction techniques, and a perpetual drive for optimal service provision incumbent on rapid diagnosis at earlier gestations, the challenges facing practitioners of fetal medicine and obstetrics are greater than ever. Even with vast improvements in ultrasound technology by the use of high-frequency transducers, transvaginal sonography, and increasing processing speed,1 not all abnormalities can be diagnosed, which can be a source of emotional distress for parents. In these instances, a further form of examination of the fetus is desirable.
Fetal magnetic resonance imaging (MRI) has evolved in the last 25 years since it was first described2 and has become an important complementary imaging modality for evaluating fetal and placental pathology. In this article, we will describe the evolution of MRI from its inception to the current day and review the indications for which MRI has been found to contribute positively to the diagnosis of fetal abnormalities, concentrating on the fetal brain.
Fetal MRI was first performed using low-strength magnets (0.08–0.35T) with T1 weighting,1 the result of which was that MRI sequences took several minutes.3 Initial attempts to visualize the fetus were hampered by fetal movement, which degrades MR images and prevents acquisition of consistent, high-quality images.4 Attempts were made to obviate this by administration of benzodiazepines to the mother or by percutaneous cordocentesis and injection of pancuronium bromide, both of which have associated risks.5–8 Advances in MRI technology have produced ultrafast T2-weighted MRI techniques, including single-shot fast spin echo that effectively freezes fetal motion without the need for pharmacological intervention.9,10 Ultrafast imaging obtains the information required for 20 slices within 20 seconds. The initial focus of MRI was on the central nervous system (CNS) because of the contrast between cerebrospinal fluid and brain tissue but, with improvement in techniques, evaluation of further organ systems, anatomy of the umbilical cord, and amniotic fluid volumes, as well as assessment of maternal structures have become feasible and clinically useful.9
It is generally accepted that MRI is safe in pregnancy and no short- or long-term effects of MRI on mother or fetus have been reported. Studies in animals of supranormal field strengths and exposure times have found no increase in the rate of teratogenic events or chromosomal deletions.9 Concerns about the safety of MRI in the fetus relate to biological effects and acoustic noise. Static field exposure has been the subject of embryonic research, but there is no conclusive evidence regarding safety1; an extensive review of the evidence concurs and recommends further work to confirm the safety of MRI.11 Guidelines have been set out by the National Radiation Protection Board on the amount of heat, termed ‘specific absorption rate’, that can be generated within the tissues of the patient while conducting the examination.10 Manufacturers currently set limits for the specific absorption rates for each pulse sequence to ensure that body temperature increase is <0.5°C.1 Although no side-effects of fetal MRI have been identified, the National Radiation Protection Board recommends performing fetal MRI starting from the second trimester of gestation. If an examination is necessary within the first trimester, the benefits should exceed the possible side-effects. The use of gadolinium-based intravenous contrast agents is not recommended, as it is known that gadolinium chelates cross the placental barrier and that delay in clearance from the fetus can potentially increase the risk of exposure to free gadolinium ions. Concerns have also been raised in some studies about the level of acoustic noise, which can reach up to 98 dB, although amniotic fluid dampens the noise received by the fetus by up to 30 dB. This attenuation of sound within the abdomen is still under investigation.10,11 Our own local data suggest that there is no detrimental effect on neonatal hearing following fetal MRI (unpublished). For up-to-date guidelines on MRI during pregnancy the reader is referred to the Health Protection Agency website (http://www.hpa.org.uk).
Fetal MRI is now typically performed using higher field strengths of 1.5T. The woman is laid on her back (or in the left lateral position to relieve the pressure on the large veins situated posteriorly in the abdomen). She is placed feet first into the scanner to reduce the claustrophobic effect that placing an individual head first into the scanner may induce, as this avoids the individual passing through the tunnel before imaging. The images take 20 second to obtain, and in an experienced centre with good equipment the total time required for the examination may only be 20 to 30 minutes. If multiple areas are being examined or there is more than one fetus, the examination may take longer.
There are standard sequences used for an in utero MRI assessment of any abnormality, and these have been described elsewhere.10
Several groups have evaluated the reproducibility and technical quality of MR images. In utero MRI of the fetal CNS is becoming part of clinical practice in several centres worldwide. However, a sound knowledge of anatomy at each gestational age is essential to the diagnosis. Comparisons between neuroanatomical data and MRI of cortical brain development are ongoing in several centres.12–14 Neuroanatomical information is also available from MRI of the preterm neonate, and postmortem studies have provided essential information on brain development. The normal neuronal migration patterns seen on anatomical specimens are also seen on in utero images. It must be remembered that, because postmortem MRI is carried out at higher resolution and with thinner slices (1–2 mm), sulcation that is seen on these images may be missed on 5 mm-thick in utero MR slices. A delay in development of up to 8 weeks (average 2wks) compared with autopsy data is reported.13 It is, therefore, essential that the radiologist interpreting the results has knowledge of normal development as seen on in utero MR images, otherwise there may be incorrect diagnosis of lissencephaly or other cortical malformation. Figure 1a–c demonstrates the changing sulcal and gyral pattern with gestational age.
Large, prospective studies are needed to assess the value of in utero MRI in the clinical setting compared with antenatal ultrasound, with follow-up studies of each fetus imaged to assess its accuracy. Follow-up studies should involve clinical assessment, pathological assessment, and imaging of the fetus/child depending on the outcome of the pregnancy. In certain cases, follow-up should continue until at least school age. An initial report that an infant is healthy post delivery does not confirm or refute a diagnosis of a brain abnormality, and for abnormalities such as isolated ventriculomegaly, for which in utero MRI is being requested in many centres, the impact may not be detectable until school age. Studies involving a multidisciplinary team are essential to obtain accurate information.
Fetal brain development is important to obstetricians and paediatricians because a detailed understanding of normal brain development allows them to understand how neurogenic processes can be disturbed at various stages of development by different aetiological factors. In addition, an understanding of normal developmental parameters is necessary for the early detection and analysis of malformations that are important in the planning of pregnancy course, delivery, and neonatal care.15–17
The fetal brain and nervous system has been the most studied organ system and occupies the largest proportion of literature devoted to MRI studies in utero, mainly as a result of the contrast between cerebrospinal fluid and neural tissue. Studies have been published that consider the virtues of MRI as a complementary modality to ultrasound,18,19 and numerous others extol the benefits of MRI as a second-line imaging investigation in the examination of both normal and pathological development of the CNS, which, in some cases, can provide useful clinical information that may alter management.16,20–24 Glenn23 has written a comprehensive review of the different stages of embryological development of the fetal brain and the structures that can be visualized using MRI.
MRI has distinct advantages over ultrasound when the position of the fetal head is low in the pelvis or is difficult to see because of maternal habitus or the presence of another fetus precludes detailed examination. In addition, a full exploration of the CNS in three spatial planes – axial, sagittal, and coronal (Fig. 2a–c) – is obtained more consistently in the second and third trimesters through MRI than with ultrasound alone.17,25,26 Fetal MRI work should be performed in centres with the most up-to-date equipment and by radiologists who have the most experience to ensure that the scans are performed and analysed to the highest possible standards. As experience in performing MRI with modern, up-to-date scanners increases, knowledge about the developing fetal nervous system will increase. Although there is disagreement about MRI findings even among those who perform fetal MRI regularly,27 with the amount of information now available the aim would be for those involved, along with input from fetal medicine specialists, to achieve a degree of consensus about the interpretation and significance of their findings. This increased exposure and experience will ultimately improve training for future radiologists, who can potentially offer fetal MRI in other units where a need has been identified.
It is well established that MRI adds valuable information and is extremely sensitive in detecting spinal anomalies including dysraphisms and spina bifida lesions.20,28,29 Myelomeningocoeles are almost always seen in association with a small posterior fossa and herniation of cerebellar tissue into the cervical subarachnoid space – the Chiari II malformation. Generally, Chiari II malformations and myelomeningocoeles are easily seen on prenatal ultrasound scans. However, MRI can be a useful adjunct in the instances previously identified, such as larger body habitus, oligohydramnios, or when the fetal spine is positioned posteriorly with respect to the mother.
Fetal MRI is also useful in identifying additional CNS anomalies that are frequently associated with spina bifida lesions, such as callosal agenesis (see section below), periventricular nodular heterotopia, cerebellar dysplasia, syringomyelia, and diastematomyelia.20
The other most common reasons for referral for fetal MRI are the presence of fetal ventriculomegaly, suspected agenesis of the corpus callosum, and posterior fossa abnormalities. More recently, MRI has been employed to study disorders of neuronal migration and midline disorders, which have always been difficult to detect sonographically. Research is ongoing as to the uses of MR spectroscopy in the prediction of fetal outcome after suspected fetal brain injury, when subtle changes in metabolites within the brain may indicate neuronal damage caused by hypoxia, as may be seen in 40% of monochorionic twin pregnancies complicated by twin–twin transfusion syndrome and death of one of the twins.25,26,30,31
The walls of the ventricles are critical to brain development and can be evaluated with fetal MRI. This germinal matrix is the innermost layer of the fetal cerebral hemisphere and is thickest earlier in gestation, before becoming a single layer of ependymal cells lining the wall of the lateral ventricle by the early third trimester.32,33 Overall, ventricular size shows good concordance with ultrasound but may show variation of 1–2 mm in the axial plane compared with the coronal plane.34,35 The cavum septum pellucidum is also visible between the lateral ventricles until birth.
There is a debate as to the utility of fetal MRI in cases of sonographically diagnosed ‘isolated’ ventriculomegaly. The rate of associated malformations, both CNS and non-CNS anomalies, is often underestimated at the time of ultrasound. Studies consistently report that about 10% of associated anomalies are recognized after birth, and the percentage of additional abnormalities diagnosed by MRI in other series varies from 5 to 50%.36 These additional findings include disorders of neuronal migration, for example lissencephaly (Fig. 3),37 holoprosencephaly (Fig. 4),24,38,39 agenesis of the corpus callosum (Fig. 5),24,40 and evidence of intraventricular and periventricular haemorrhage (Fig. 6).38,39 Conversely, MRI can be used to confirm the absence of mild asymmetrical ventriculomegaly (>2 mm difference between ventricle size) detected on ultrasound, which, although rare, has been found to be associated with a significant incidence of neurodevelopmental delay.40–43 The findings of MRI thus have bearing on counselling at the time of finding an anomaly and have the potential to alter management based on their severity.19 Mild ventriculomegaly, defined as an atrial width of 10–15 mm when isolated, is associated with an 11–16% risk of developmental delay according to a review of numerous studies investigating outcomes in mild ventriculomegaly.43 Several investigators have suggested that an atrial width of 10–11.9 mm is associated with a better outcome than a width of 12–15 mm. However, a recent thorough, systematic review did not suggest a difference in abnormal neurological outcome in these two groups.43 Rutherford44 identifies the difficulties in interpreting imaging findings such as unilateral ventriculomegaly and understanding their clinical significance, and imaging siblings and parents may aid in this process. She highlights the challenge of ascertaining the clinical significance of abnormal findings based on newer imaging techniques and how lack of this knowledge can affect the way in which parents are counselled, can contribute to their stress without adding to their care, and can affect their decision-making.
The appearance of the fetal supratentorial brain on MRI has been well studied, and there are tables of normal values for size and development at each stage of gestation from 22 weeks gestational age23; several benchbooks are now available. The cerebral mantle has a multilayered appearance until 28 weeks’ gestation, consistent with different developing layers of the brain (Fig. 7). The appearances in vivo of the differential dissolution of the subplate zone show a good correlation with histological and postmortem MRI specimens, which indicate that the subplate persists longer in the frontal and anterior temporal lobes – areas of higher cortical function.45 Studies have shown that diffusion-weighted imaging is sensitive to the normal maturational changes that occur with advancing gestation and suggest that this form of imaging may be sensitive to abnormalities of brain development that may not be apparent on T2- or T1-weighted images, but this is still under investigation.32
Agenesis of the corpus callosum can be found in between 0.03% and 0.7% of mid-trimester ultrasound scans46 and may exist in isolation or be a marker for other abnormalities in the brain. At least 46 malformation syndromes and metabolic disorders have been reported in patients with complete agenesis of the corpus callosum or hypoplasia (dysgenesis).46 The corpus callosum is best viewed on 3 mm midline sagittal images, on which it is seen in its entirety at the superior margins of the lateral ventricles, is uniform in thickness, and increases in length with gestation (Fig. 5).32 A series of coronal images is also helpful. Pistorius et al.19 argue that, in studies in which MRI was found to be superior to ultrasound in diagnosing partial or complete absence of the corpus callosum, this probably reflected poorly performed studies rather than an inherent disadvantage of ultrasound. In studies that employed a mid-sagittal view, ultrasound was found to be as good47,48 or better than MRI49 for diagnosis of corpus callosum abnormalities between 21 and 34 weeks.
Although callosal agenesis has been reported as an incidental or isolated finding, most patients manifest varying neurological symptoms, including developmental delay, cognitive impairment, and epilepsy.50 In one series of 29 cases, 23 were found to have delayed or abnormal cortical morphology, and 15 were found to have cerebellar or brainstem morphology or both. Those surviving with a poor outcome had either infratentorial pathology or abnormal cortical sulcation in addition to the agenesis.51 Doherty et al.52 published a wide range of outcomes in 189 individuals. Affected sufferers were more likely to have hydrocephalus, microcephaly, or other phenotypic abnormalities. They were also more likely than their siblings to suffer seizure disorders, cerebral palsy, autism, developmental delay (80%), and learning disorders. Difficulties with feeding, visual and hearing problems, and alteration in pain perception were also commonly reported. As stated previously, ultrasound scans and MRI are complementary modalities in accurately identifying agenesis of the corpus callosum and any associated abnormalities. If these are seen, parents should be informed that most of those affected will have neurodevelopmental problems and that the condition could be part of a more complex syndrome. Counselling can be more difficult if no additional problems are identified, as the outcome can be variable. As a result, termination of pregnancy is a valid option in either instance.46
Neuronal migration disorders are relatively common causes of congenital brain malformations. They can broadly be divided into four main groups. In the first group, neurons either do not migrate at all from the ventricles (periventricular heterotopia) or migrate only halfway (subcortical band heterotopia). In the second group, some neurons. but not all, reach the cortex. No normal cortical layers are formed (lissencephaly, pachygyria, cobblestone cortex). In the third group of disorders, neurons overshoot the cortex and end up in the subarachnoid space (marginal-leptomeningeal glioneuronal heterotopia, cobblestone cortex). The fourth group of disorders result from disruption of the late stage of migration and cortical organization (polymicrogyria).
Most migration disorders have a genetic basis, and many genes have been discovered in recent years. The same gene mutation can cause an overlap between the genotype and the phenotype because (1) mutations affect protein function differently and (2) the effect of a given mutation may be moderated by somatic mosaicism or, in the case of X-linked genes, by skewed X inactivation. For a full review on the genetics of neuronal migration disorders, which is outside the scope of this article, see Barkovich et al.15 and Guerini et al.53
The subtleties and various forms that neuronal migration disorders take predispose them to be missed on imaging, especially on ultrasound.46 However, it has previously been shown that there is a correlation between findings on MRI and ultrasound that helps to determine the extent and importance of the lesions. MRI has a definite advantage over ultrasound in its ability to assess normal brain sulcation and gyration40,54 (Fig. 1) and is the only modality that can evaluate myelination and neuronal migration.13,55,56 The optimum time to perform the examination is between 30 and 32 weeks. If scanning is performed earlier, it should be repeated at this gestational age.57 The appearance of the sulci follows an organized and spatial temporal pattern.12,58 Examination of sulcation in pathological specimens is considered to be one of the most accurate ways of dating a fetus.59 Owing to the resolution of MRI, the appearance of the sulci seems to lag behind fetal pathology specimens by an average of 2 weeks (see Table I), displays right to left asymmetry, and may be delayed in twin gestations.12,13,57,58
|Sulcus||Detectable by pathology in 25–50% of fetuses (wks)||Detectable by fetal magnetic resonance imaging in 75% of fetuses (wks)|
The Sylvian fissures are the first sulci to appear and can be seen before 18 weeks’ gestation. By 34 weeks, all primary and some secondary sulci are visible on fetal MRI.60 Because of their precise appearance, knowledge of the correct gestational age at the time of MRI is critical in interpreting the sulcation pattern.32 If there are any doubts, then the scan should be repeated 4–6 weeks later. Fetal MRI has been reported to be superior to ultrasound in identifying schizencephaly (Fig. 8), lissencephaly (Fig. 3), polymicrogyria, and grey matter heterotopias in a small study of 20 fetuses with the diagnosis confirmed on either postnatal imaging or autopsy.57,61 Lissencephaly and polymicrogyria are generally associated with severe psychomotor retardation and intractable seizures.53 Some forms of polymicrogyria have a genetic basis associated with various known genes.53 Other forms of polymicrogyria are acquired and are caused by disruptions, for example fetal cytomegalovirus infection and prenatal hypoxic–ischaemic encephalopathy, including that due to vascular problems related to twinning. Polymicrogyria is often seen at the borders of porencephalic cysts and schizencephaly defects. Schizencephaly is generally considered to represent a form of neuronal migration disorder, but is more likely to be diagnosed prenatally than other forms of the general disorder because of the disruption in cerebral architecture.46 The incidence is approximately 1.5 per 100 000 births62 and occurs in the presence of cortical clefts, in either or both cerebral hemispheres, lined with grey matter extending from the plial surface of the brain into the ventricle. As indicated, there may be adjacent polymicrogyria. It is usually sporadic, but occasional familial cases arising from developmentally expressed genes in the homeobox region have been reported. As with polymicrogyria, cytomegalovirus infection and vascular disruption62 have been reported to cause the condition sporadically.46 There is thus an overlap between dysgenetic and encephaloclastic processes in the brain that can give rise to similar findings.
Myelination occurs predominantly after birth, but the protein and lipid signals of the white matter sheath can be seen on MRI from 30 weeks onwards.19 Failure of myelination may be a sign of a metabolic defect but may not be seen until after birth owing to maternal compensation for the defect. Poor myelination associated with other parenchymal pathology, for example in pyruvate dehydrogenase deficiency, or with cysts and impaired cortical development, as would be seen in Zellweger disease, may be morphologically identified by MRI prenatally.63 Confirmation of these diseases requires invasive testing by means of chorionic villus sampling and is usually undertaken when the fetus is at increased risk because of a family history of the disease.19 Infants born with neuronal migration disorders usually experience convulsions, which are often difficult to control, and exhibit severe neurodevelopmental disabilities. The early involvement of a paediatric neurologist to advise on anticonvulsant medication is important.46
MRI plays a role in the diagnosis of posterior fossa abnormalities including mega-cisterna magna, Dandy–Walker malformation (Fig. 9) and its variant form, vermian agenesis/hypoplasia, cerebellar hypoplasia and atrophy, as well as pontine abnormalities.
Cerebellar agenesis is rare and usually occurs in association with chromosomal disorders (most commonly trisomies 13 and 18). It is associated with severe motor disabilities, including cerebral palsy. Cerebellar hypoplasia may be part of a recessive or syndromic disorder and results in more variable disabilities, most involving ataxia, mild to moderate learning difficulties, or dyspraxia.46 Vermian agenesis, as seen in the Dandy–Walker, malformation can be associated with severe neurological deficits and additional developmental malformations of the brain (such as agenesis of the corpus callosum and neuronal migration disorders) and of other organs. Some patients, however, can retain relatively normal intelligence. Vermian agenesis can be diagnosed by ultrasound scans, but in one study the sonographic diagnosis was not supported by the autopsy findings in more than 50% of cases.64 These figures would tend to support the use of an additional modality in the form of MRI to clarify cases in which the diagnosis may not be certain.
However, although for some disorders the evidence is overwhelmingly in favour of MRI over ultrasound, there are conflicting reports about the value of MRI in accurately identifying defects in the posterior fossa. MRI can identify the position of the cerebellar tentorium and integrity of the cerebellum regarding its anatomy and biometry.65 If a finding of increased retrocerebellar fluid-filled space is isolated, the outlook is likely to be good. However, before the third trimester, inferior vermian agenesis can be overlooked and cerebellar hypoplasia may be difficult to identify.21,66 Pontine hypoplasia and a ‘kinked brainstem’ (Fig. 10) may be seen with MRI, but more detailed visualization of the brainstem is more difficult.67 In a recent retrospective review concentrating on fetal posterior fossa abnormalities there was complete agreement between fetal and postnatal MRI in only 59% of cases. In 15%, the postnatal scan excluded the fetal MRI diagnosis, and, in 26%, additional anomalies were revealed.68 This again demonstrates important issues associated with the introduction of a new technique and illustrates the need for combining data from antenatal ultrasound and MRI with postnatal assessment of the neonate by MRI.44
The causes of fetal ischaemia can be placental (infarction or infection), fetal (infection), or maternal (hypovolaemic shock, hypoxia, abdominal trauma, hypertension) in origin. Fetal MRI can be used to assess ischaemia involving the cortex or the white matter. Ischaemia can be demonstrated shortly after the insult in the third trimester by means of a hyperintense diffusion-weighted image signal. Chronic changes are more commonly seen and appear as increased signal on T2-weighted MR images and decreased signal on T1-weighted MR images because of subcortical leucomalacia.69 An insult occurring before 21 weeks results in a porencephalic cavity (Fig. 11), and after 26 weeks there is an intense astrocytic proliferation and a septated cavity with irregular walls. The later the insult, the less likely that a cystic component will occur.57 Monochorionic twin pregnancies are complicated by increases in both congenital malformations and acquired brain injury after death of one of the twins. This is associated with significant morbidity and mortality of the surviving twin (Fig. 12).25,26,30 One MRI study of a series of monochorionic twins, of whom one died in the second trimester, found changes, including polymicrogyria, subependymal cysts, intracranial haemorrhage, ventriculomegaly, and delayed sulcation, in 33%.70 In most of these fetuses, ultrasound findings were normal, and thus these changes would not have been detected before birth, reinforcing the need to use MRI as a complementary modality to detect potentially subtle changes secondary to an ischaemic insult. It should, however, be noted that the authors themselves advised treating their findings with caution as there was a lack of correlation with neurodevelopmental outcome. One child who was found to have polymicrogyria associated with a hypoxic–ischaemic lesion on MRI has congenital hemiplegia and epilepsy, but the full scale of the relationship between subtle fetal MRI findings such as delayed sulcation and neurodevelopmental outcome is not known. Thus, further work correlating MRI findings with outcome is essential before any conclusive statements or accurate counselling can be offered for future patients and their unborn infants.
Spontaneous CNS haemorrhage (Fig. 13) may result from congenital or drug-induced coagulation disorders or fetal compromise. It may also occur within a pre-existing tumour or vascular malformation (e.g. vein of Galen aneurysm) and from intrauterine interventions such as laser coagulation of arteriovenous connections in twin–twin transfusion syndrome. Intraventricular haemorrhage may result in an obstructive hydrocephalus or arrest normal brain development and neuronal migration.9
Primary infection with cytomegalovirus can result in a 30–40% chance of transmission to the fetus, causing structural and profound neurodevelopmental abnormalities (Fig. 14), particularly if transmission occurs in the first half of pregnancy.44 Anomalies that can be detected are microcephaly, ventriculomegaly, and porencephaly, which can be detected antenatally by ultrasound and MRI. More subtle signs include borderline ventriculomegaly with focal calcifications, best seen with ultrasound, and small lateral periventricular cysts that are symmetrically situated.71 Congenital herpes simplex infection can lead to severe parenchymal destruction, which can be seen on MRI. Diffusion-weighted imaging might be useful for the detection and monitoring of brain abscesses and herpetic encephalopathy from 35 weeks onward.72
Fetal MRI has advanced rapidly in the last 25 years and has progressed from being a research tool to an integral part of the fetal diagnostic process. Its ability to discern more complex structural defects, which may predict functional neurological disability in some instances, and its use in the abnormally sited fetus or the mother with a raised body mass index is well recognized. The large investment in equipment and specialized training to perform and interpret data limits the extent to which MRI can be employed. Its use should be focused in specialized centres with the most experience and modern equipment, and to this end it will never replace ultrasound as the modality of choice in initial screening and diagnosis of structural abnormalities. Nevertheless, at an undoubtedly difficult and stressful time for anxious parents, MRI has added to the growing amount of information that can be imparted during counselling and has established itself as a vital adjunct to our armoury of investigations in prenatal care.