When first used in the mid-1980s, magnetic resonance (MR) image acquisition was slow, necessitating maternal sedation and even temporary fetal paralysis with intramuscular injection to limit motion artifact. The last few decades have brought many technological advances allowing very fast MR acquisitions. Vendor-specific protocols are widely available and have been used to effectively image the fetus. Images are acquired in 1 second or less, eliminating the need for sedation. Motion artifact, although still present, is significantly reduced. MR has better soft tissue contrast than ultrasound, can image a larger field of view allowing a global perspective of the fetus, and is not hindered by bony interfaces, maternal obesity, oligohydramnios, or an engaged fetal head. Referral for fetal MR evaluation is most frequently encountered in the setting of suspected fetal central nervous system (CNS), thoracic, genitourinary, or gastrointestinal disorders. Maternal indications include evaluation of abdominal and pelvic pain, suspected pelvic masses, or placental invasion.

A commentary by the American College of Obstetrics and Gynecology was published summarizing the National Institutes of Health Workshop on fetal imaging technology, including present and future fetal MR applications, which are discussed in the following section.¹

A primary concern of referring physicians and their patients is the safety of both mother and fetus. MR utilizes no ionizing radiation, but questions arise regarding fluctuating electromagnetic fields, high sound intensity levels, and tissue heating.

Several animal and tissue studies have been conducted to determine the biologic effects of electromagnetic fields. An early study looked at the long-term effects of repetitive exposure to a static 1.5 tesla (T) magnetic field on human lung fibroblasts.² Study and control group proliferation was similar, indicating no adverse effect from repetitive MR exposure. Human studies are rare. A large epidemiologic retrospective study compared spontaneous abortion rates, infertility rates, incidence of low birth weight, and premature delivery among nurses and technologists working with MR before and after employment.³ There were no increased incidences of adverse outcomes in the MR-exposed group. Two studies reported follow-up on children exposed to echo-planar MR in utero. The first by Baker et al completed a 3-year follow-up on 20 children with no demonstrable increase in the occurrence of disease or disability.⁴ These patients were imaged with a 0.5 T superconductive magnet from 21 weeks to term. In the second case-controlled prospective observational study of 20 infants exposed to echo-planar imaging, pediatric assessments completed on these infants at 9 months of age were normal.⁵

In reference to the sound intensity, one study assessed the sound level experienced by the fetal ear during an MR procedure by having a volunteer swallow a microphone connected to a thin lead and filling the stomach with a liter of fluid to represent an amniotic sac.⁶ There was at least a 30-decibel (dB) attenuation of intensity from the body surface to within the fluid-filled stomach, which reduced the acoustic sound pressure down from the dangerous threshold of 120 dB to an acceptable level less than 90 dB. This level is lower than the 135 dB experienced when vibroacoustic stimulation is used. No evidence of hearing loss was found in one study after 450 babies were exposed to this noise intensity from vibroacoustic stimulation.⁷ The potential for tissue heating during MR evaluation was evaluated using a pig model. No significant temperature alteration in pig fetuses with intramniotic, brain, and abdominal cavity temperature probes was detected using an ultrafast MR imaging sequence (HASTE).⁸ Studies have also looked at fetal heart rate patterns during MR procedures and reported no change in heart rate patterns or incidence of fetal movements.⁹

Recommendations from a recent document from the American College of Radiology concerning safe practices for MR have determined that MR can be performed in pregnancy if the data are needed to affect the care of the fetus or mother during pregnancy.¹⁰ Furthermore, because of the potential for dissociation of the chelate molecule in the amniotic fluid, gadolinium-based MR contrast agents should not be routinely administered. The longer the chelate molecule remains in a protected space such as the amniotic sac, the greater the potential for dissociation of the potentially toxic gadolinium ion. It is also recommended that women undergoing MR imaging during pregnancy provide informed written consent documenting that they understand the potential risks and benefits of the procedure, namely, that there are no known risks to mother or fetus from MR imaging during pregnancy regardless of trimester, but exhaustive long-term studies have not been done, and in theory potential risk may still exist. If, however, the information gained by imaging will impact management of mother or baby during the pregnancy, and the information cannot be obtained by nonionizing means (ultrasound), the benefit may outweigh any small or theoretical risk of the procedure.

All fetal and maternal studies are performed on a 3T magnet or less. The magnet should only be used within clinical parameters and follow FDA guidelines for nonsignificant risk devices. In addition, all women complete a written MR safety screening questionnaire, including information regarding metallic implants, pacemakers, or other metal- or iron-containing devices that may impact the study.¹¹ Maternal iron supplementation may rarely cause artifact in the colon but does not usually affect the fetal evaluation.

In over 2500 MR procedures performed at our institution, maternal anxiety secondary to claustrophobia and/or fear of the MR equipment has occurred in less than 1% of our patients. To reduce maternal anxiety in the small group of women unable to proceed with the outpatient exam, we recommend either oral Valium (5-10 mg) or Ativan (1-2 mg) for sedation to the ordering obstetrician, recognizing that these medications are pregnancy class D and that the patient will need to arrange a ride home after the examination. Otherwise, anxiolytics are not routinely employed. When maternal obesity is an issue, recognize the typical MR tube opening (gantry) is 60 to 70 cm, and therefore the capacity of the gantry is often exceeded prior to the reported table weight limit. When patients are over 300 pounds, depending on where the weight is most prevalent, it may not be possible to optimally position the gravid abdomen in the center of the magnet.

Women are placed in the supine or left lateral decubitus position. A torso coil is used in most circumstances, with the occasional use of either the body or cardiac coil depending on maternal size, fetal size, and area of interest. A series of three-plane localizers are obtained relative to the maternal coronal, sagittal, and axial planes. In every fetal case, the gravid uterus is imaged in the maternal axial plane (7-mm slices, 0 gap) with a T2-weighted fast acquisition, typically a single-shot fast spin echo sequence (SSFSE), half-Fourier acquisition single-shot turbo spin echo (HASTE), or rapid acquisition with relaxation enhancement (RARE) equivalent depending on the brand of machine and manufacturer (Tables 48-1 and 48-2). These acquisitions are particularly good in identifying the fetal and maternal anatomy. Next, a fast T1-weighted acquisition such as spoiled gradient echo (SPGR) is performed (7-mm thickness, 0 gap). The pertinent fetal and maternal anatomic parameters identified on these sequences are noted on the MR patient report.

Orthogonal images of targeted fetal or maternal structures are then obtained. In these cases, 3- to 5-mm slice thickness, 0 gap, T2-weighted acquisitions are performed in the coronal, sagittal, and axial planes (Figure 48-1). Depending on the anatomy and underlying suspected abnormality, breath hold T1-weighted images can be performed to evaluate for subacute hemorrhage, fat, or location of normal structures that appear bright on these sequences, such as liver and meconium in the colon (Figure 48-2).^12,13 Steady-state MR imaging, such as steady-state free precession (SSFP) sequences are relatively motion insensitive with high signal-to-noise ratio (SNR), which is helpful in both fetal and placental evaluation. Short tau or short inversion-time inversion recovery images (STIR) may provide contrast between the abnormal and normal structures when there is only a subtle difference in water content between the tissues, as frequent in the case of thoracic masses such as congenital pulmonary airway malformation (CPAM) or sequestration to normal lung. Diffusion-weighted imaging (DWI) is helpful in brain evaluation and may be employed to evaluate the placenta.^12,13

Our exams include an axial brain 3- to 5-mm T2-weighted sequence to obtain head biometry for the purposes of assigning an approximate gestational age from the biparietal diameter and head circumference in all cases (see Figure 48-1A).¹⁴ The thickness of the slice should be applied based on the size of the fetus, size of the area of interest, and consideration of the trade-off between the potential of improved conspicuity with thinner slices and the decrease in the SNR.

Some MR nomograms of the head have been proposed, which include only the brain and none of the bony calvarium as a better indicator of actual brain volume.¹⁵ For the purpose of gestational age assignment, however, we prefer the more standard inclusion of the skull, as defined by historic sonographic nomograms.^16,17 The brain itself can be measured when there is an abnormality and suspected parenchymal underdevelopment or loss.

The most common indication of the fetal MR evaluation is for suspected central nervous system abnormalities on sonographic evaluation and accounts for approximately 70% of our practice. MR images of the fetal CNS are hypothetically superior to that of ultrasound because of the improved visualization and tissue contrast. Ultrasonography remains the modality of choice for initial identification of CNS abnormalities. However, MR is a very useful adjunct in the antenatal diagnosis of some suspected CNS anomalies. Fast T2-weighted images produce excellent tissue contrast, and cerebrospinal fluid–containing structures are bright (Figures 48-3 and 48-4). This allows exquisite detail of the posterior fossa, midline structures, and cortex. Near-field attenuation secondary to the fetal skull on ultrasonography is not encountered with MR.

Another major advantage of MR is the ability to acquire images in the axial, coronal, and sagittal planes in reference to the fetus or the maternal pelvis. Sagittal and coronal fetal images are very helpful, for example, in evaluating the vermis and corpus callosum (Figures 48-5 and 48-6). T1-weighted images are used to assess for hemorrhage and can also differentiate fat from fluid containing structures (Figure 48-7).

Similar to sonographic scans, CNS biometry can be measured and is routinely done in our cases. Routine measurements include biparietal diameter, occipital frontal distance, cerebellar width, cisterna magna depth, and bilateral atrial measurements.^14,18,19 Ventricle and cisterna magna measurements on 60 fetuses beyond 14 weeks to term gestation found atrial measurements to be slightly smaller on MR compared with ultrasonography, but determined the abnormal sonographic cutoff of greater than 10 mm was greater than two SD above the mean on MR (Figure 48-8).¹⁸ Cisterna magna measurements were gestational age dependent with cutoff values less than 4 and greater than 11, also similar to ultrasonography. Nomograms for multiple components of brain biometry, including the corpus callosum length and cerebellar vermis, have also been published.^15,20 The measurements of corpus callosum and vermis lengths are routinely done on all MR studies of the central nervous system (see Figure 48-1C).

Cortical maturation can be evaluated by MR images, and cerebral gyration and sulcation patterns reflect the embryologic development.^21,22 Fetuses with a CNS abnormality may have significant lag time in cortical development.²¹ Ultrasonography is very limited in evaluating very subtle migrational abnormalities at early gestational age. These migrational CNS disorders such as Walker-Warburg are problematic, but MR is becoming more reliable in assessing for cortical maturation, especially later in gestation (Figure 48-9).^22-24

Four large studies assessed the second-opinion MR examination in the setting of a suspected CNS abnormality demonstrated on ultrasonography in regard to its ability to confirm, change the diagnosis, and possibly alter clinical management. In the first study with 66 cases of abnormalities found on confirmatory sonogram, MR added additional information in 57.6% of cases, and changed the diagnosis in 40% of cases.²⁵ Clinical management was clearly changed in nine cases. In 73 MR examinations performed for CNS anomalies, another group found that 48% of pregnancies were managed differently than the way they would have been managed if the diagnosis had been based only on a sonographic basis.²⁶ Additional information was provided in 46 of 72 pregnancies (64%), and the diagnosis was changed in 20 of those 46 cases. Clinical management was altered in eight cases. MR was more likely to confirm sonographic diagnosis prior to 24 weeks, but beyond 24 weeks a change of diagnosis or additional information was gleaned more frequently.²⁷ A British study confirmed the findings of the other studies, where MR imaging either changed the diagnosis or gave additional information that altered management in 35 of 100 cases.²⁸

Presently, the most common reason for fetal MR referral is isolated ventriculomegaly on ultrasound, which may be associated with a highly variable outcome depending on other abnormalities present.^29-32 Additional abnormalities commonly associated with ventriculomegaly may not be defined on sonographic examination (Figure 48-10). The most common change in diagnosis was from marked ventriculomegaly to a more precise diagnostic category, such as aqueductal stenosis or hydrancephaly (Figure 48-11) and from mild ventriculomegaly to more specific diagnoses, including agenesis of the corpus callosum and migrational abnormalities (see Figure 48-9). Primitive abnormalities may be better defined with MR (Figures 48-12 and 48-13). A more exact diagnosis impacts patient counseling and, to a lesser degree, clinical management.^27-33 This finding was confirmed in a recent paper, which found that MR imaging provided important information in the setting of ventriculomegaly, especially in fetuses with other findings of the central nervous system by sonographic evaluation.³¹ At the present time, there is considerable variability in the central nervous system diagnosis in the setting of ventriculomegaly secondary to modality differences (ultrasonography or MR) and observational errors. Agreement was 60% with sonography and 53% with MR in this series.³³