Magnetic resonance imaging examination of the fetal brain

Abstract

Smith et al. reported the first use of magnetic resonance imaging (MRI) in pregnancy in the Lancet in 19831. At that time, they concluded: ‘we believe that it will prove to be safe and (…) the ability to study tissue water will give it wide application in obstetric practice’. More than 20 years later, we must acknowledge that they were right; a PubMed search (in October 2006) on ‘fetal MRI’ found more than 1500 references, of which 700 were on ‘fetal brain MRI’. This illustrates how MRI has rapidly evolved from being a tool used for highly targeted examinations in select populations of fetuses, to one that is widely used for prenatal diagnosis2, most commonly within the fetal brain. Fetal brain MRI is available in many prenatal diagnostic centers and radiologists have developed highly specialized skills to perform and interpret such examinations. However, because of the rapid development of this methodology, there may be confusion regarding the information that a standard fetal brain MRI examination should provide. The aim of this paper is to provide tips and a framework for performing fetal brain MRI examination. Ultrasound examination is the method of choice for routine screening in the fetus and for examination of fetal brain anatomy. Even when MRI is available, findings on brain sonography should be considered carefully, particularly when focusing on the fontanelle3, 4. In certain situations, especially when using T2-weighted images only, and when the ultrasound examination is performed by a highly trained operator and using the transvaginal route, MRI fails to achieve better results than sonography4. However, even in expert hands, some abnormalities may be overlooked by ultrasound due to poor local conditions (e.g. maternal obesity, inappropriate fetal position, decreased amniotic fluid) or technical difficulties (e.g. near-field reverberation artifact, which makes sonographic evaluation of the proximal ventricle difficult) or because the anomaly is very subtle. Such cases benefit from MRI, and previous studies have demonstrated that it may have advantages over ultrasound in selected cases5-7. In particular, some aspects of brain maturation and myelination are better assessed with MRI8, 9, which has proved useful in the diagnosis of various fetal brain abnormalities2, 10. Nevertheless, it would be unwise to overestimate the complementary contribution of MRI, which should always be performed only after an optimal ultrasound examination by a skilled sonographer. A history of severe brain abnormality in a previous pregnancy, but ultrasound examination is considered normal; MRI is performed in order to look for subtle signs of recurrence. An abnormality identified on ultrasound examination that appears to be isolated (typically ventriculomegaly or corpus callosum agenesis); MRI is performed to look for potential abnormalities that may have been overlooked by ultrasound. An abnormality diagnosed on ultrasound examination, but the examination cannot be completed because of technical problems (e.g. maternal adiposity, fetal position); MRI is performed to supplement ultrasound. A high risk of development of brain abnormality, especially in cases of fetal infection (mostly cytomegalovirus, varicella and toxoplasmosis) or ischemic damage (in-utero death of a monochorionic twin, twin-to-twin transfusion syndrome)10; in some centers, MRI is routinely offered in such cases. The timing of fetal MRI can be influenced by a number of factors, for example, local legislation that regulates elective termination of pregnancy (TOP). In countries such as France, the UK or Israel, TOP is permitted up until term11. Although MRI can be performed from the early second trimester, examination of the fetal brain this early in gestation is not recommended since it does not allow the best possible examination: some sulci are not visible before 28 weeks, while narrow pericerebral spaces hinder visibility of the sulci after 34 weeks10. We prefer to wait until the beginning of the third trimester (28–32 weeks) to maximize information gathered from fetal brain analysis10. However, parental anxiety and medical necessity, as well as ultrasound findings, may lead to the examination being performed at an earlier gestational age. Precise identification of abnormalities early in gestation is not easy12, and some abnormalities may develop late in pregnancy, such as white matter abnormalities related to cytomegalovirus infection during the third trimester13. If MRI is performed during the second trimester, the amniotic fluid is still abundant, the fetal head is not engaged in the maternal pelvis and fetal movements can generate artifacts. Therefore, the optimal gestational age should be decided for each case individually, according to the risks and benefits of this diagnostic procedure, the nature of the abnormality, and the legal age limit for TOP where the scan is being performed. Structural/anatomical malformations (such as corpus callosum agenesis or abnormalities of the posterior fossa) may be assessed early, whereas evaluation of gyration and the parenchyma require waiting until well into the third trimester. After informed consent from the mother has been obtained, fetal MRI should be performed using a high-field MR scanner (1.5 T) in order to obtain optimal imaging results. A phased-array surface coil should be used. However, the signal may decrease if the studied structure lies too far away from the coil. This may happen in multiple pregnancies, obese women or when the woman is in the lateral decubitus position14. Fetal sedation may be achieved by maternal oral administration of 1 mg flunitrazepam15 20 min before the examination; this benzodiazepine is used for sedation and anxiety reduction. Although fast sequences do not always require sedation16, all sequences should be performed at least during maternal breathholding. MRI has been used in pregnancy for more than 20 years17, 18. Many studies have reported that it is not associated with any major side effects and should be considered harmless19-22. The Safety Committee of the Society for Magnetic Resonance Imaging concluded 15 years ago that prenatal MRI was indicated when other non-ionizing diagnostic imaging methods were inadequate or when MR examination would provide important information that would otherwise require the use of ionizing radiation23. Elevation of fetal temperature is potentially hazardous, but it is difficult to ascertain the degree of heating of the fetus during an MRI scan24. No significant temperature rise was observed in pregnant pigs when imaged with half-Fourier acquisition single-shot turbo spin-echo (HASTE) sequence on a 1.5 MRI unit25. Another major concern is the acoustic noise generated by MRI; there are currently no data regarding the level of acoustic noise observed in the fetus following in-utero attenuation of sounds24. The entire fetal brain volume should be explored using the three spatial planes. Fetal MR examination should start with a localizer in three planes, in order to further orientate the examination. The main aim is that it should allow for all subsequent scans to be oriented along fetal rather than maternal axes. This convention should be followed in all cases, both in clinical practice and for research purposes10, 16. According to the interval between successive excitation pulses (repetition time), the time at which the signal is recorded (echo time) and the possible application of a gradient reversal, several sequences can be used26. Changes in these different parameters generate changes in tissue contrast. Fast T2-weighted sequences remain the basis of fetal MRI and are the most frequently used. Technical advances now allow fast acquisitions (< 30 s), such as in single-shot fast spin echo (SSFSE) or based on half-Fourier acquisition (with a different acronym according to the manufacturer, e.g. HASTE (Siemens); RARE (Brucker)). Such fast acquisition sequences provide good resolution of fetal tissues, particularly the brain because of its high water content. They are used for surface delineation, sulcation analysis and brain biometry10. Figure 1 illustrates sagittal and coronal T2-weighted images, on which biometric measurements are performed. Small, hypointense lesions, such as tubers, small calcified leukomalacia and myelination, are not easily detected with T2-weighted sequences10, 26-29, thus emphasizing the need for a combination of sequences, with at least some T1-weighted images. Thick-slab T2-weighted sequences generate a three-dimensional impression of the brain26-28, 30. Midline sagittal (a) and coronal (b) T2-weighted magnetic resonance images, showing biometric measurements. In the midline sagittal slice (a), the fronto-occipital diameter is measured as the distance between the frontal and occipital lobes (long arrow). Height and anteroposterior diameter of the cerebellar vermis are measured on this same slice (small arrows). In the coronal slice (b), biparietal diameters (BPDs) are measured at the level of the temporal horns. The bone BPD corresponds to the distance between the two internal tables of the skull, while the cerebral BPD is the measurement of the greatest transverse diameter of the brain. Gradient-echo T2-weighted images (T2*) are accurate in the detection of chronic hemorrhagic lesions and calcifications10, because hemosiderin deposits or calcifications create a magnetic dipole that is responsible for a loss of signal, well-depicted on sequences with a long echo time (TE) (Figure 2). T2* images are not sufficient to provide a detailed analysis of the fetal cerebral parenchyma. Contribution of gradient-echo T2-weighted magnetic resonance imaging in the diagnosis of old hemorrhages. On T2-weighted imaging, the hemorrhage produces a faint hyposignal (arrow) anterior to the right frontal horn in the axial view (a), whereas on gradient-echo T2-weighted imaging, the hyposignal (arrow) is marked (b). Although fast gradient-echo sequences are used in T1-weighted imaging, the acquisition time (> 1 min) may require fetal sedation to minimize motion artifacts31. T1- weighted sequences are less efficient than are T2-weighted ones for assessing brain limits, but they may help in the detection of lesions that can be overlooked using T2-weighted sequences, such as laminar necrosis and calcified leukomalacia10. Figure 3 illustrates a case in which calcified leukomalacia was detected easily on the T1-weighted image (Figure 3a), while being overlooked on the corresponding T2-weighted image (Figure 3b). It has also proved useful for detecting hemorrhage32 3–4 days after it occurs, when methemoglobin is formed by oxidative denaturation of hemoglobin and its paramagnetic effects are responsible for a marked T1 hyperintensity. Lipomas, tuberous sclerosis nodules10, 33 as well as the process of myelination are better visualized using T1-weighted sequences31. Coronal T1-weighted slice (a) demonstrating a calcified leukomalacia lesion (arrow) at 33 weeks (that was due to toxoplasmosis) appearing as a focal hyperintensity in the white matter external to the right atrium. The lesion was overlooked by T2-weighted imaging (b). Diffusion-weighted MRI (DWI) (Figure 4) reflects the Brownian motion of water molecules within the brain and can be evaluated by measuring the apparent diffusion coefficient (ADC). It is necessary to calculate ADC because hyperintensity in diffusion-weighted images can be attributed to T2 hyperintensity (T2 shine-through) and therefore decreased diffusion may be overlooked when examining diffusion-weighted images only. Diffusion tensor imaging provides information on the direction of the white matter tracts (anisotropy) and depends on the tissue structure and the direction of the applied gradient34. It has provided valuable information in adults, mainly in the early diagnosis of strokes35. It has also proved valuable after birth for detecting the early stages of hypoxic–ischemic damage and various white-matter disorders36. In fetuses, measurement of the ADC might be useful in the future to detect ischemic lesions or to assess brain maturation37, 38. Moreover, diffusion anisotropy characterizes premyelinating structures39, thus allowing identification of the unmyelinated fibers of the corpus callosum. Hemimegalencephaly with diffusion abnormalities. (a) T2-weighted axial slice showing a typical pattern of hemimegalencephaly with enlarged right cerebral hemisphere, right ventriculomegaly and opercular dysplasia. The right hemisphere is hypointense compared with the left one. (b) Diffusion-weighted imaging with apparent diffusion coefficient mapping. The diffusion coefficient is decreased in the right cerebral hemisphere. Practically, the potential applications of DWI are three-fold: (i) assessment of fiber development by diffusion tensor imaging, for example in the vicinity of gyration abnormalities or in association with corpus callosum agenesis; (ii) assessment of the myelination process through monitoring DWI changes with brain maturation, based on ADC measurement and mapping37; and (iii) early detection of lesions leading to cerebral ischemia, prior to the establishment of irreversible damage. This is the main challenge. FLAIR (fluid-attenuated inversion-recovery) sequences might prove useful in cases of hemorrhage40 as this sequence creates more contrast to structures adjacent to corticospinal fluid spaces14. Investigators have also reported abnormal findings in preterm infants with evidence of periventricular white-matter injury41. Perfusion imaging of the fetal brain using spin-tagging techniques, the BOLD (blood oxygenation level-dependent) method or even contrast injection42-44 may develop in the next few years. Although several clinical and toxicological studies in animals have suggested that contrast agents may be harmless during pregnancy, their use is not recommended in order to avoid potential deleterious effects to the fetus45. Gadolinium chelates have been shown to cross the placenta43, 46; they appear in the fetal bladder and are thus reabsorbed from the amniotic fluid by fetal swallowing45. Their biologicical half-life in the fetus is not known. More recently, several authors have raised concerns about the potential toxicity of gadolinium for patients with renal compromise (nephrogenic systemic fibrosis)47. Magnetic resonance spectroscopy is a non-invasive tool for the examination of fetal cerebral metabolism48. It is based on the measurement of chemical shift, i.e. the local change in resonant frequency due to different chemical environments. Resonances of various metabolites are identified in order to obtain MR spectra. Normal patterns and normal variants are still not fully determined. It seems necessary to establish a metabolic mapping of the fetal brain at different gestational ages. When normal patterns are established, the purpose of this technique will be the early detection of metabolic variations in fetuses with suspected brain damage. It may become possible to diagnose some inborn errors of metabolism with magnetic resonance spectroscopy34, 49. Finally, fetal functional imaging may prove informative in the next few years50; local neuronal activity increases in response to a vibroacoustic or a visual stimulus and is responsible for a different MR signal, thus providing information about the activity of a cerebral area. The correct interpretation of fetal brain MRI requires extensive training. As with ultrasound examination, an adequate MRI examination should be performed according to a predefined approach, as described below. Biometry is an important and easy initial step in fetal brain assessment by MRI. Normal reference ranges from large cohort studies are not yet available, but we report the main landmarks for normality in Table 1. The fronto-occipital and cerebral biparietal diameters are important because they allow evaluation of the fetal brain volume. In fact, unlike ultrasound examination, only the brain is measured, and so the result is not hampered by variability in pericerebral spaces. Bone biparietal diameter can be measured and its difference from cerebral biparietal diameter, gives an evaluation of the pericerebral spaces. The length of the corpus callosum can be measured on a sagittal plane from the genu to the posterior extremity of the splenium (Figure 1). Unlike ultrasound examination, MRI allows both lateral ventricles to be measured on coronal planes at the level of the atria51. There is good agreement between MRI and ultrasound measurements51. However, the choice of the best threshold to define ventriculomegaly remains controversial52. Cerebellar measurements (Table 1) should include assessment of the vermis by measuring its height, anteroposterior diameter and surface area; the transverse cerebellar diameter should also be measured10. Other measurements such as the width of the third ventricle and anteroposterior diameter of the fourth ventricle, interhemispheric distance and width of the lateral sulcus can also be performed, but these are not used in current practice10. The second step in MRI examination of the fetal brain should be assessment of gyration53, which is a good marker of fetal maturation. At 27 gestational weeks, the interhemispheric fissure, lateral sulcus and internal parieto-occipital, hippocampal, callosal, calcarine, cingulate and central sulci should already be present10, 54 (Figures 5 and 6). At 29 weeks, marginal, pre- and post-central, intraparietal, collateral and superior temporal and frontal sulci should be visible and the central sulcus should reach half of the cerebral hemisphere in depth10, 54, 55 (Figures 5 and 6). At 31 weeks, the inferior frontal sulcus should be visible. Finally, at 35 weeks, the temporal lobe should have all its sulci, including superoanterior, inferior and external occipitotemporal sulci. At this gestational age, the gyration should have obtained its definitive pattern10, 54, 55 (Figures 5 and 6). T2-weighted coronal images at 25 (a), 30 (b) and 35 (c) weeks. At 25 weeks (a), the secondary sulci are not yet visible. At 30 weeks (b), the superior frontal sulcus (arrowhead), inferior frontal sulcus (large arrow) and superior temporal sulcus (small arrow) are visible. At 35 weeks (c), again note the superior temporal sulcus (arrowhead). The inferior temporal sulcus (small arrow), collateral sulcus (checked arrow) and external occipitotemporal sulcus (large arrow) are also visible. The gyration of the external aspect of the hemispheres is well developed. Sagittal magnetic resonance images at 25 (a), 30 (b) and 35 (c) weeks. At 25 weeks (a), secondary sulci are not yet visible. At 30 weeks (b), the cingulate sulcus (arrowhead), marginal sulcus (large arrow) and calcarina fissure (small arrow) are visible. At 35 weeks (c), note again the cingulate sulcus (arrowhead), marginal sulcus (large arrow) and calcarina fissure (small arrow). The internal parieto-occipital fissure (checked arrow) is also visible and appears much thinner than it did at 30 weeks. Secondary cingulate sulci are also visible, accounting for the tortuous appearance of the cingulate sulcus. The gyration of the external aspect of the hemispheres is well developed. Like gyration, myelination is a good indicator of fetal cerebral maturation. It is thought that the increase in cholesterol and glycolipids accompanying the formation of myelin results in an increase in bound water, thus leading to a shortening of T1 and T2 sequences, visible as a hypersignal on T1-weighted images or a hyposignal on T2-weighted images. Such signal changes can be seen in the white matter, starting at 20 weeks' gestation in the posterior brainstem. At 27 weeks, some myelin is visible at the level of the vermis and the middle cerebellar peduncles. A moderate signal is also visible at the central basal ganglia10, 54. Up to 33 weeks, the myelination of the posterior limbs of the internal capsules occurs, extending progressively to the globus pallidus at 35–36 weeks. Figure 7 illustrates this myelination process. It can also be detected along the optic tracts55. Axial T1-weighted slices showing myelination of the posterior limbs of the internal capsule (arrows). At 25 weeks (a), there is no evidence of myelination, while at 35 weeks (b), there is marked T1 hyperintensity, reflecting the myelination process. Here we evaluate the diagnostic contribution of MRI with respect to different brain areas: midline, periventricular, cerebral parenchyma, cerebral surface, pericerebral spaces, ventricles and posterior fossa. In the majority of cases, the optic chiasma, the olfactory bulbs and the pituitary gland cannot be evaluated by ultrasound, and the detection of abnormalities in these regions relies on the information that can be provided by MRI. While the diagnosis of callosal agenesis is based on ultrasound, MRI has proven useful in the investigation of partial agenesis10, 56 (Figure 8), as well as for the identification of any associated structural abnormalities that are likely to impact on the prognosis10, 56, 57. Other abnormalities of the corpus callosum, such as hyperplasia or lipoma (Figure 9), might be amenable to MRI diagnosis10. Diffusion tensor imaging might prove useful as an additional tool in the assessment of abnormalities involving cerebral fiber development. T2-weighted midline sagittal slice showing partial agenesis of the corpus callosum (arrow). The splenium is missing. Interhemispheric lipoma with dysgenesis of the corpus callosum. (a) T2-weighted midline sagittal slice. The interhemispheric lipoma is hypointense (white arrow). The anterior part of the corpus callosum is not depicted clearly, while the posterior part appears very thin (black arrow). (b) T1-weighted midline sagittal slice without fat suppression. The lipoma is hyperintense (white arrow). (c) T1-weighted midline sagittal slice with fat suppression. The lipoma is not visible. In other midline pathologies, such as holoprosencephaly or absence of the septum pellucidum, ultrasound examination performs sufficiently well to provide the correct diagnosis, but MRI might help in the diagnosis of lobar holoprosencephaly or in looking for abnormalities associated with septal agenesis10. The diagnosis of septo-optic dysplasia, however, is still extremely difficult during fetal life58, because hypoplasia of the optic nerves and/or chiasma is very difficult to assess even with MRI and may develop only after birth. One of the greatest advantages of MRI is the detection of subependymal heterotopias; the indented outline of the ventricles is better visualized with MRI than it is with ultrasound59. Figure 10 gives an example of heterotopia. While MRI is much better than ultrasound at detecting subependymal nodules in tuberous sclerosis (Figure 11), a normal fetal MRI cannot rule out this diagnosis33. Germinolysis is usually well depicted by ultrasound. T2-weighted coronal slice at 32 weeks showing left periventricular heterotopia (arrow) indenting the external border of the atrium. This was diagnosed on magnetic resonance imaging performed because of sonographically isolated mild ventriculomegaly. Parenchymous tuber in a 36-week fetus presenting cardiac rhabdomyomas. It appears on this T1-weighted axial slice as a hyperintense lesion (arrow) in the right centrum semiovale. Tuberous sclerosis was confirmed after delivery. Abnormalities of the cerebral parenchyma may be focal or diffuse. Most commonly, lesions are not developmental but are acquired in-utero10. MRI allows detection of small intraparenchymal hemorrhages which may be overlooked by ultrasound (Figure 2)60-64, and offers good visualization of the whole cerebral parenchyma. However, it is mandatory to use T1-weighted sequences for this purpose, although T2* gradient-echo echoplanar sequences may also be helpful. Diagnosis of ischemic lesions may also benefit from MRI (Figure 12)16, 31: cavitary lesions are T1 hypointense and T2 hyperintense, whereas laminar necrosis and periventricular calcified leukomalacia are T1 hyperintense65, emphasizing once again the need for using both types of sequence. The latter lesions can be hyperechoic but their echogenicity is variable and they can be missed by ultrasound. Gliosis and white-matter edema appear as diffuse T2-hyperintense and T1-hypointense lesions (Figure 12). Usually, the more diffuse the lesion, the harder the diagnosis. DWI and spectroscopy are certainly useful in the detection of some diffuse parenchymal abnormalities65, 66. Different patterns of cerebral ischemia. (a) Gliosis appears on this T2-weighted axial slice as diffuse white-matter hyperintensity. Cavitations (subependymal pseudocysts) are visible along the lateral ventricles (arrows) and are T2-hyperintense. (b) Calcified leukomalacia appears on this T1-weighted coronal slice as hyperintensities (arrows) above the lateral ventricles. Abnormal cell differentiation, as seen in tuberous sclerosis (tuber may be located in the parenchyma (Figure 11)) or hemimegalencephaly (Figure 4) should also benefit from MRI examination2, 10, 12, 33, 67. So far, only isolated cases of fetal cerebral tumors imaged by MRI68-71 have been reported and no study has analyzed systematically the contribution of fetal MRI in such cases. The timing of the appearance of the different sulci has been established by MRI57, emphasizing that MRI must be performed sufficiently late in pregnancy to detect abnormal gyration patterns72. Visualization of sulcation with ultrasound is possible73 and lissencephaly can be diagnosed sonographically. However, abnormalities of cell proliferation, as seen in microlissencephaly (or microcephaly with simplified gyration pattern), are better detected with MRI (Figure 13)10, 12. MRI can detect subtle abnormalities such as polymicrogyria, corresponding to an overfolding of the cortical ribbon and may contribute to the diagnosis of schizencephaly, which presents a diagnostic challenge10, 74, 75. Figure 14 illustrates a case of polymicrogyria associated with schizencephaly and septal agenesis. Laminar necrosis (Figure 15) can also be diagnosed by MRI10, and the technique is useful in the evaluation of macrocephaly10. T2-weighted coronal slice at 31 weeks showing micrencephaly with simplified gyral pattern (a). The superior frontal sulcus (1) and the collateral sulcus (2) are less visible compared with the gestational age-matched control (b). The superior temporal sulcus is visible in the control (3) but not in the fetus with microlissencephaly. T2-weighted coronal slices showing polymicrogyria associated with schizencephaly and septal agenesis in a 33-week fetus. In (a), note the right open lip schizencephaly (thin arrow) with dysplastic cortex (large arrow) lining the cleft (polymicrogyria confirmed at pathological examination). In (b), note the left closed lip schizencephaly (thin arrow) with dysplastic cortex (large arrow) below the cleft and the septal agenesis (star). Laminar necrosis: T1-weighted axial slice showing abnormal opercularization of the Sylvian fissures, with marked T1 hyperintensity of the cortex (arrow) along the right Sylvian fissure due to laminar necrosis. When enlarged, pericerebral spaces are visible on ultrasound, although they are usually only partially visible. Some lesions, such as subdural hematoma, may be missed by ultrasound, but are amenable to MRI diagnosis (Figure 16) because pericerebral spaces are always visible with MRI10. Bilateral subdural hematoma. This fetus was referred for magnetic resonance imaging (MRI) because of macrocrania and mild bilateral ventricular dilatation on ultrasound. The pericerebral space is enlarged and presents a double component: there is slight hypointensity in the subarachnoid space (white arrow) and hyperintensity in the subdural space (black arrow). In the postnatal period, MRI showed a bilateral subdural hematoma. There is extensive literature on the use of MRI in the diagnosis of ventriculomegaly. Although the optimal sonographic measurement at which MRI should be offered remains controversial52, assessment of ventriculomegaly should benefit from its use, particularly in determining the presence of any associated structural abnormality and for studying the periventricular white matter76. MRI can also confirm a hemorrhagic origin of ventriculomegaly, even several weeks after the acute episode of bleeding (Figure 17). Ventricular dilatation related to intraventricular hemorrhage. (a) T2 gradient-echo axial slice at the level of the lateral ventricles. The right ventricle is dilated. The marked hypointense ring (arrow) around this ventricle is due to the old intraventricular hemorrhage. (b) T1-weighted axial slice at the same level. The small hyperintense focus (arrow) visible along the lateral aspect of the right ventricle indicates a semi-recent hemorrhage; the old hemorrhage is not visible on this sequence. Imaging the fetal posterior fossa is challenging because many abnormalities discovered during pregnancy that involve the posterior fossa have a poor prognosis. Imaging the fetal brain stem with ultrasound is theoretically possible77, but in common practice, ultrasound cannot consistently provide good visualization of this structure, while MRI can. Figure 18 gives an example of brain stem hypoplasia. T2-weighted midline sagittal slice showing brain stem hypoplasia at 27 weeks. The bulge of the pons is present but the whole brain stem is too thin. Hypoplasia was confirmed by neurofetopathological findings. Diagnosing partial vermian agenesis (Figure 19) is also challenging and ultrasound and MRI can be used as complementary tools78, 79. Analysis of the vermian area may contribute to the evaluation of prognosis in Dandy–Walker malformation; the prognosis is much poorer when the vermis is severely dysplastic compared with when there is partial vermian agenesis80. T2-weighted midline sagittal slice showing partial posteroinferior vermian agenesis at 29 weeks. The fourth ventricle is open. The posterior lobe (below the primary fissure (arrow) is not twice the size of the anterior lobe (above the primary fissure) as it is in the normal brain. MRI has rapidly evolved from being a highly targeted examination applied to a very select population of fetuses into a more widely used technique for prenatal screening and diagnosis of fetal brain anomalies. It has proved useful in the evaluation of most fetal brain anomalies and is likely to play a greater role in the coming years. It should be kept in mind that MRI must always be performed following a comprehensive ultrasound examination. Because it is shared between many different prenatal specialists, such as radiologists, obstetricians, neuropediatricians and geneticists, particular effort should be made to standardize the diagnostic approach to include biometric, gyral and parenchymal assessment.