Chapter 44: OBSTETRICAL APPLICATIONS FOR 3D ULTRASONOGRAPHY

Three-dimensional ultrasonography (3DUS) can be combined with conventional two-dimensional ultrasonography (2DUS) for the evaluation of normal and abnormal fetal anatomy. Development of cost-effective high-performance computers and sophisticated image analysis software now makes it practical for 3DUS to be conveniently integrated with standard diagnostic imaging equipment. The previous chapter summarized basic concepts regarding the acquisition and fundamental concepts of ultrasound volume data. This overview will describe how this information can be applied to specific problems during pregnancy.

Although 3DUS applications have been described for over 2 decades, one of our initial challenges was to evaluate their clinical value against a backdrop of emerging technology. In 2005, a literature review of more than 500 related articles suggested that additional diagnostic information was particularly useful for facial anomalies, neural tube defects, and skeletal malformations.¹ Other potential benefits involved early pregnancy evaluation, nuchal translucency measurements, weight estimation, fetal lung volumetry, growth evaluation, and possibly maternal–fetal bonding. Technical developments, such as spatiotemporal image correlation (STIC) algorithms, were also cited for improving our ability to examine rapid changes in the fetal heart. Chapter 45 will more closely examine the emerging field of four-dimensional ultrasonography.

We have witnessed an impressive refinement of image analysis tools and their migration from dedicated image workstations to notebook computers. This technology permits the examiner to choose among many different visualization techniques with a variety of output display modes as summarized in Chapter 43. In most cases, volume sonography can be generally considered as an important diagnostic technique that is complementary to conventional 2D imaging. Before specific examples of how 3DUS can be applied to pregnant women are reviewed, we will first describe a few general concepts for how 3DUS can be best applied during pregnancy.

Use 2D Ultrasonography for an Initial Diagnostic Impression

Health care professionals are classically trained to translate 2D images into volume reconstructions for an improved understanding of spatial relationships and image patterns that suggest congenital abnormalities. Volume sonography provides the examiner with an opportunity to systematically evaluate anatomic structures with less interobserver dependency, sometimes in ways that are not possible through conventional 2D methods. A preliminary 2D sonographic scan for suspected fetal abnormalities can be very useful, and the results can be used to guide the subsequent 3D study. Once an initial differential diagnosis has been generated, the examiner should be able to formulate pending questions about the initial diagnostic impression that may or may not be satisfactorily addressed using 3DUS. If spina bifida is detected, for example, there may be a need to more precisely define the anatomic level of the defect. This information would help the clinician to estimate the risk of problems related to ambulation and bowel or bladder function.

In order to better understand the contribution of conventional imaging to 3DUS, Gonçalves et al² compared both modalities in a blinded manner for 99 fetuses at a mean menstrual age of 24.4 ± 6.5 weeks. Complete agreement between 2DUS and 3DUS was observed for 90.4% of the findings. Six anomalies were missed by volume sonography, including ventricular septal defects (n = 2), interrupted inferior vena cava with azygous continuation (n = 1), tetralogy of Fallot (n = 1), and cystic adenomatoid malformation (n = 1). There were 12 discordant diagnoses between 2D and 3D ultrasonography. When compared to postnatal diagnoses (n = 106), the sensitivity and specificity of volume sonography was 92.2% and 76.4%, respectively. However, no significant differences in the diagnostic conformity to neonatal outcomes were found when the 3DUS results were compared to the sensitivity (96.1%) and specificity (72.7%) from 2DUS (P = .23). This study concluded that the information provided by 2D ultrasonography is consistent, in most cases, with the diagnostic results from 3DUS.

Match the 3DUS Technique With a Specific Clinical Question That Is Being Asked

The diagnostic benefit of volume sonography critically depends upon expectations of the examiner and the precise reason why the scan itself is being performed. Although most patients relate to this technology from its ability to display a fetal face for bonding purposes, this is not a recognized medical indication in low-risk pregnancies. In this context, one must recognize that 3DUS has the potential of revealing much more than a fetal face. The fundamental technologies, as described in the preceding chapter, actually represents several different image visualization tools that may or may not apply to the question being asked. They include multiplanar imaging, surface rendering, volume rendering, thick-slice scanning, tomographic slices, inversion mode, and 3D Doppler ultrasonography (Figures 44-1,44-2,44-3,44-4,44-5,44-6, and 44-7). As one example, the maximum intensity projection algorithm might be the most appropriate to analyze a possible hemivertebra that is suspected on the 2D scan. Complementary display modes, such as multiplanar imaging or the use of parallel tomographic slices, might also offer benefit. By comparison, a surface-rendering algorithm would not be expected to provide useful information for bony structures such as cranial sutures or spinal abnormalities. This modality would be better suited for delineation of soft tissue problems such as cleft lip. Some volume analysis software packages allow the examination of volume reconstructions using a combination of different rendering algorithms (Figure 44-8). Others may use 4DUS for evaluating the status of joint movement of the wrist in a fetus suspected of having arthrogryposis. Hence, the examiner must remember to formulate one or more specific questions that are based on the initial 2D scan in order to choose the most appropriate type of image analysis.

Do Not Underestimate the Diagnostic Value of 3D Multiplanar Imaging

Although much emphasis has been placed on volume rendering, one must appreciate how useful it is for an examiner to analyze multiplanar image slices from standardized orthogonal views. In our practices, it is not uncommon that we might use this modality as the initial method of fetal organ evaluation. After standardized orthogonal planes are displayed, it is possible to systematically "march" through a volume of interest in order to better characterize the anatomy. One of the most common applications for this modality will be demonstrated for the fetal face evaluation. From these views, it will also be possible to make a wide variety of quantitative distance, angle, and area measurements as needed.

Good Two-Dimensional Imaging Is Likely to Translate to Satisfactory Three-Dimensional Volume Acquisitions

The most useful volume data sets will be acquired when optimal sonographic technique is used in pregnancies where technical factors do not degrade the image. Remember the basics by adjusting scanning parameters such as system gain and acoustic focus. A reasonable goal is to adjust the region of interest in the rendering box so that it fills about two-thirds of the available display screen. Scan from different viewing perspectives for optimal images of the specific region of interest. For example, the fetal spine and overlying skin line should be obtained from a prone fetus, not from the supine position.

We are often asked, "Is the sonographer's job at risk if all one needs to do is to place a transducer on the maternal abdomen and push a start button?" The answer is emphatically, "NO!" because the likelihood that volume sonography will provide an adequate means for diagnostic interpretation is really dependent on the original quality of the images. The examiner should be able to optimize the image by utilizing the proper probe position while making certain that excessive pressure is not placed on the maternal abdomen. There are several ways to analyze good-quality volume data sets if the original acquisition is obtained under optimal conditions. Careful attention should be paid to adjusting image depth, acoustic focus, and signal gain at a time when fetal movement is minimal for acquisition of this image data. Both 2D and 3D ultrasonographic techniques, however, are sensitive to the potentially detrimental effects of maternal obesity, fetal movement, early fetal age, presence of maternal abdominal scars, and oligohydramnios.

With these caveats in mind, some of the more common diagnostic applications will now be reviewed to further demonstrate basic approaches for using volume sonography during pregnancy.

Early Fetal Anatomic Evaluation

The first trimester 2D scan is mainly performed for dating early pregnancies, confirming cardiac activity, and to measure nuchal translucency as part of risk assessment for aneuploidy. Relatively fewer investigators have used conventional 2D imaging techniques to evaluate anatomic structures of the developing embryo. An important advantage of accurate early diagnosis would include parental reassurance that a major abnormality was not present. Earlier anomaly detection would provide more time for defining pregnancy options, prenatal care, delivery plans, and prognosis. As the field of fetal therapy develops, the earlier diagnosis of certain types of anomalies may even allow treatment at a time when chances for salvaging the fetus may be higher at an earlier stage of disease progression.

Several investigators have used 2D endovaginal scans to visualize fetal anatomy before 14 weeks' menstrual age.^{3,4, and 5} First trimester pregnancies can also be evaluated using 3DUS, although recent improvements in high-frequency probe technology have made it increasingly feasible to visualize cross-sectional and surface anatomy of the embryo (Figure 44-9). Blaas et al⁶ were among the first groups to use 3DUS for volume reconstruction of embryos between 7 and 10 weeks' menstrual age. They used a customized 7.5 MHz annular array vaginal probe and a VingMed system to trace contours of embryonic brain structures that permitted volume calculations as well.

Normal embryonic anatomy has also been studied using 3DUS.⁷ One investigator threaded a catheter-based 20-MHz ultrasound transducer through the maternal cervix and endometrial cavity to demonstrate the embryonic face, limbs, and brain.⁸ Benoit et al⁹ introduced the term, "sonoembryology" for describing the specific use of volume sonography of the developing fetus. He later demonstrated the evolving changes in embryonic brain development and reported that the optimal time to use 3DUS was between 7 and 12 weeks.¹⁰ Twelve representative images of the rendered volume provided an informative timeline of embryonic brain development (Figure 44-10). There have also been several related reports of conjoined twins,^{11,12, and 13} spina bifida,¹⁴ holoprosencephaly,^{10,15,16,17, and 18} sirenomelia,¹⁹ and sacrococcygeal teratoma.²⁰

Although the sensitivity of 2DUS for the detection of major fetal anomalies between 11 and 14 weeks has been reported to be approximately 50%,²¹ there are many factors that can affect diagnostic results. They include the population being studied, type of equipment used, examiner experience, exam duration, and the gestational age at which the scan is performed. Unfortunately, no large-scale comparative studies have been performed that can clearly define the role of 3DUS for the systematic detection of fetal anomalies during early pregnancy as compared to 2DUS. Technical improvements, such as the development of more sophisticated image analysis tools or the commercial availability of higher-frequency transducers, are likely to result in more accurate and earlier diagnoses over time. The best diagnostic technology in the world, however, will not be translated into improved outcomes unless clinical management protocols are also capable of taking advantage of this information.

How much additional benefit is actually provided by 3DUS over conventional 2DUS during early pregnancy? Michailidis et al²² compared real-time 2D scans (5-MHz transabdominal and 7-MHz transvaginal probes) to 3DUS (7-MHz transvaginal probe) for early fetal anatomic assessment. The anatomic survey included the head, face, stomach, abdominal wall, kidneys, bladder, spine, and extremities in 159 consecutive women at 12.0 to 13.9 weeks' gestation. Two 3D volumes were obtained at the end of the procedure, and they were subsequently analyzed for the same anatomic features that were examined using 2DUS. A "complete" anatomic survey was possible in 93.7% of the 2D scans, which was a higher visualization rate than 80.5% of fetuses that were satisfactorily examined using 3DUS (P < .001). The mean time to perform a 2D scan was 12.2 ± 3.4 minutes (SD) as compared to the greater time it took to acquire and analyze 3D volume data sets for the same information (8.4 ± 1.5 minutes) (P < .001). They concluded that real-time 2DUS was the best way to examine first trimester embryonic anatomy. Volume sonography provided occasional views that were not possible using 2DUS, was associated with less scanning time, and provided a mechanism where scanned data could be stored for subsequent review.

Fauchon et al²³ recently performed a prospective study where an examiner acquired transabdominal 3DUS of the entire fetus from 273 singleton pregnancies between 11.0 and 13.9 weeks' menstrual age. Each data set was manipulated and analyzed by 2 independent examiners who were blinded to each other's results. The requirements for sonographic visualization of 12 anatomic structures were strictly defined. Crown-rump length and nuchal translucency were measured with a high degree of agreement between both investigators in 100% and 84.6% of cases, respectively. Negligible clinical differences in either measurement resulted among all 3 examiners. In this study, a single abdominal 3DUS acquisition of a fetus between 11 and 13 weeks' gestation usually provided satisfactory views for both the anatomic survey and nuchal translucency measurement of the embryo. Increasing maternal weight was an important factor that prevented adequate visualization of embryonic anatomy, whereas longer crown-rump length increased the odds of the examiners being able to identify half of these anatomic characteristics.

The aforementioned studies indicate that the entire fetus can usually be satisfactorily screened for an anatomic survey during early pregnancy. Factors such as embryonic position, bladder filling, examiner experience, type of equipment used, menstrual age, limited range of probe movement, transducer frequency, and even maternal obesity can greatly influence the examiner's ability to complete this transvaginal study. The advantages for using a high-frequency transvaginal probe to improve image resolution is offset by relatively limited tissue penetration. It also becomes increasingly difficult to capture the entire fetus after approximately 14 weeks' menstrual age. After this time, it may be necessary to acquire more than one volume data set due to the technical specifications of most mechanical volume probes. Accordingly, one could subsequently attempt volume acquisitions with a lower frequency abdominal probe. More studies are required to characterize the normal appearance and evolution of embryonic structures on the basis of 3DUS. For example, the physiologic mid-gut herniation and its relationship to development of ventral wall defects is well known. This normal finding is typically very prominent in a 10-week fetus (see Figure 44-9).

Genetic Risk Assessment

Several sonographic markers of genetic risk have included an evaluation of nuchal translucency (NT), nasal bone, frontomaxillary facial angle, fetal pelvic measurements, and embryonic volume. Key questions must be considered before these new techniques are introduced into clinical practice. First, the technique must be relatively simple to perform, affordable, and reproducible among different examiners. Second, the relationship of these measurements to menstrual age must be established. Third, the new marker should be able to distinguish between normal and pathologic pregnancies with good sensitivity and a relatively low false-positive rate. Fourth, any potential advantages over the use of conventional 2DUS and any technical limitations of the new technique should be described.

Nuchal Translucency

At least 6 reports have examined how 3DUS can be used to evaluate nuchal translucency.^{24,25,26,27,28, and 29} Paul et al²⁷ studied 40 consecutive uncomplicated pregnancies that underwent first- trimester screening for Down syndrome at 11 to 14 weeks. Nuchal translucency was measured using both transabdominal 2DUS and 3DUS. Two volume data sets were acquired—one that included a midsagittal plane and a second one that was obtained from a random initial plane. This detail is important because images that are analyzed from the original plane of acquisition are usually the clearest because they do not require reconstruction. Small but significant errors can occur if the nuchal translucency thickness is in the range of the ultrasound beam's lateral resolution of 1 to 2 mm. They found that nuchal translucency thickness could be repeated in 38 of 40 (95%) of the volumes that were acquired from a sagittal view of the fetus. By contrast, the random volumes yielded only 24 of 40 embryos with satisfactory nuchal translucency measurements. The mean difference between results from 2DUS and those obtained from reslicing "sagittal" volume data sets was −0.097 mm (95% limits of agreement from −0.481 to 0.675) and 0.225 mm (95% limits of agreement from −0.369 to 0.819) when random volumes were analyzed. Their results underscored the fact that the analysis of 3D volume data sets can be reliably used to replicate nuchal translucency measurements only when the nuchal skin line is also seen on the 2D scan. If visualization of the posterior neck is obscured by acoustic shadowing, one would not expect a well-defined nuchal translucency on the multiplanar reconstruction.

Clementschitsch et al²⁸ also prospectively examined 229 unselected pregnancies to compare the use of 2DUS and 3DUS for NT measurements. Satisfactory nuchal translucency measurements were obtained using 2DUS in 96.8% of cases as compared to 3DUS (98.6% transabdominal). Suboptimal fetal position was the main reason why 2DUS failed to provide satisfactory measurements. In some cases, it was difficult to precisely distinguish between fetal skin and amnion or uterine wall (6.3% for 2DUS, 3.3% for 3DUS). Fetal movement was the main reason for measurement failure using 3DUS. The mean time for either method was similar (9 minutes for 2DUS versus 10 minutes for 3DUS). Finally, the correlation between these measurements was very high (r = 0.97).

Both studies suggest that reliable nuchal translucency measurements are feasible with transabdominal 3DUS in the 11- to 14-week embryo, especially if the original plane of volume acquisition contains a midsagittal view of the embryonic neck. Of course, it would be prudent to follow technical guidelines for 2DUS that are recommended by the Foundation for Fetal Medicine (http://www.fetalmedicine.com/downs/nuchal.htm) or the Nuchal Translucency Quality Review (NTQR; http://www.ntqr.org) programs to ensure adequate quality control of these precise measurements. Presently, only the latter program addresses the use of 3DUS for NT measurement. The NTQR guidelines suggest that volume data acquisitions for this purpose are most reliable from a sagittal sweep of the embryonic face. Examiners are also cautioned about the potential limitations of lateral resolution for their ultrasound system.

Can reliable NT measurements be obtained from volume data that are not originally acquired from a sagittal plane? Shipp et al³⁰ addressed this question by analyzing NT measurements of 29 consecutive fetuses between 11.4 and 13.9 weeks' menstrual age. The 2DUS results were compared to 3D-based measurements from a median sagittal plane of the embryonic neck. A sagittal plane was obtained by navigating through 3D multiplanar views using volume data sets that had been initially acquired from a coronal sweep of the embryo. The mean measurement (SD) using 2DUS (1.7 ± 1.4 mm) was not statistically different from use of a reconstructed sagittal plane (1.8 ± 1.6 mm) (P = .4) and these results were highly correlated (r = 0.98, P < .001). Their findings suggest that nuchal translucency measurements are feasible from 3D data, despite that the original volume was initially acquired from a coronal plane. They also concurred with an earlier report that emphasized the importance of satisfactorily visualizing the neck region using 2DUS before initiating the volume sweep procedure.²⁷ The interpretation and practical implementation of their findings warrant further investigation in a larger group of patients. Furthermore, the study did not consider the skill of the examiner for satisfactory manipulation of these volume data sets in a reproducible manner.

Nasal Bone Assessment

The fetal nasal bone also has been found to be a sonographic marker of genetic risk. Cicero et al³¹ used conventional 2DUS to identify the absence or presence of nasal bone in fetuses with trisomy 21 at 11 to 13⁺⁶ weeks' gestation. The nasal bone was absent in 113 (0.6%) of the 20,165 chromosomally or phenotypically normal fetuses and in 87 (62.1%) of the 140 fetuses with Down syndrome. The inclusion of the nasal bone in first trimester combined screening for trisomy 21 achieved a detection rate of 90% for a false-positive rate of 2.5%. Another 2DUS study³² observed a much lower rate of absent nasal bone in an unselected (16.7%) and selected (46.7%) population of fetuses between 11 and 13.9 weeks' menstrual age. Because this study did not utilize any formal training or quality assurance program for the detection of nasal bone, Sonek et al³³ recommended that an ultrasound marker, such as the nasal bone, should not be used in a screening program unless the examiners are adequately trained. In an earlier study, Cicero et al³⁴ analyzed the ability of 15 sonographers for obtaining satisfactory views of the nasal bone. They found that approximately 80 2D examinations were required to achieve competency for nasal bone assessment during the routine 11- to 14-week scan. In this context, Malone et al³⁵ performed nasal bone imaging as a screening tool for aneuploidy in 6324 of 38,189 patients who were scanned at 10.4 to 13.9 weeks' menstrual age. An acceptable nasal bone image was reported in 76% of cases. Nasal bones were present in 4779 (99.5%) of this subgroup and absent in 22 fetuses (0.5%). Absence of nasal bones had sensitivity for aneuploidy of only 7.7% with a false-positive rate of 0.3% and positive predictive value of 4.5%. They concluded that nasal bone evaluation was not useful for population screening for trisomy 21, possibly because of the difficulty in performing this assessment consistently in a general U.S. population setting. However, one-quarter of these sonographers reported unsatisfactory visualization of the nasal bone despite specific training in first trimester sonography. These results seemed contrary to their specific implementation of a quality control program to monitor ongoing performance of nasal bone sonography. Nonetheless, it is presently unclear as to why the FASTER trial results in the United States did not confirm the utility of nasal bone screening in the United Kingdom.

Does 3DUS provide additional diagnostic benefit for nasal bone evaluation during early to mid-pregnancy? Rembouskos et al³⁶ found that a midsagittal view was the best original plane of 3D volume acquisition, and their results suggested that satisfactory visualization of the nasal bone was optimal when the fetal profile was insonated at about 45 degrees. The likelihood of an adequately visualized nasal bone from a 3D volume data set was strongly related to the quality of the initial 2D image. Other investigators^37,38 have used 3DUS to document the presence of a gap between the nasal bones during early pregnancy. Peralta et al³⁸ used 2DUS and 3DUS to scan 450 fetuses between 11.0 and 13.9 weeks' menstrual age. They found sonographic evidence of this gap in about 20% of fetuses. Furthermore, in about 40% of these cases, the nasal bone may erroneously be interpreted to be absent in an optimal median sagittal plane.

Gonçalves et al³⁹ used 3DUS to evaluate nasal bones in 26 fetuses with Down syndrome during the second trimester of pregnancy (Figure 44-11). Rendered facial profile views demonstrated absent nasal bones in 18.9% of cases, of which 90% had Down syndrome for an overall sensitivity of 34.6% and false-positive rate of 3.7%. This appearance was associated with a 9.3-fold increased risk for Down syndrome when compared to the normal control group. By comparison, 3 ossification patterns were demonstrated from coronal views of the rendered face: (1) normally developed, (2) delayed ossification, and (3) absent nasal bones. Sensitivity, false-positive rate, and likelihood ratio of absent nasal bones for detecting Down syndrome were 34.6%, 3.7%, and 9.0 (95% CI = 1.3 to 68.7), respectively. These investigators identified nasal bones with delayed ossification by using this maximum-intensity projection algorithm. Similar to findings from 2DUS, the absence of nasal bones was associated with the highest risk of Down syndrome. Delayed nasal bone ossification patterns were associated with a somewhat lower risk for these abnormal fetuses. The sensitivity, false-positive rate, and likelihood ratio of delayed ossification for detecting Down syndrome were 42.3%, 22%, and 1.83 (95% CI = 0.8 to 4.4), respectively. This "hypoplasic" pattern probably reflects shortened nasal bones that have been described using 2DUS.⁴⁰ Benoit and Chaoui⁴¹ subsequently used the 3D maximum intensity projection algorithm to demonstrate unilateral absence or hypoplasia of nasal bones during the second trimester of pregnancy. An analysis of these nasal bone patterns may improve our ability to identify fetuses at risk for Down syndrome. However, more experience with these methods is required in an unselected patient population before the diagnostic significance of 3DUS can be established for this purpose.

Frontomaxillary Facial Angle

Frontomaxillary facial angle (FMF) is another sonographic marker of fetal aneuploidy that has been studied in both the first and second trimesters of pregnancy. This measurement is defined as the angle between the upper surface of the upper palate and the frontal bone from a median sagittal view of the fetal face (Figures 44-12 and 44-13). Sonek et al⁴² used 3DUS to standardize this measurement, based on the hypothesis that the maxilla is dorsally displaced in relation to the forehead in fetuses with trisomy 21 between 11.0 and 13.9 weeks' menstrual age. The FMF angle was significantly larger in the abnormal group of 100 fetuses with Down syndrome (mean 88.7, range 75.4 to 104 degrees) as compared to 300 chromosomally normal controls (mean 78.1, range 66.6 to 89.5 degrees). This angle was also not significantly associated with NT. Subsequent work by the same group⁴³ underscored the importance of standardizing the FMF measurement using 3D multiplanar views of the facial profile that includes the tip of the nose and the rectangular shape of the maxillary bone. This approach was used to demonstrate very reproducible results and they were able to prospectively demonstrate that experienced examiners can alternatively measure the FMF angle using 2DUS as well.⁴⁴ Furthermore, an increased FMF angle in fetuses at 11.0 to 13.9 weeks' menstrual age is increased in cases of trisomy 13, but only when associated holoprosencephaly is present.⁴⁵ A prospective study of 782 euploid fetuses and 108 fetuses with Down syndrome combined the FMF with biochemical screening tests.⁴⁶ The inclusion of FMF angle to first trimester combined screening increased the estimated detection rate from 90% to 94% for a false-positive rate of 5%.

Molina et al⁴⁷ also applied the FMF angle to 150 normal fetuses and 23 fetuses with Down syndrome between 16 and 24 weeks' menstrual age. In the normal group, the FMF angle did not change with menstrual age and the 95th centile was 88.5. By comparison, the FMF angle was greater than 88.5 degrees in 65.2% of abnormal fetuses, and interobserver analysis indicated that in 95% of cases, the difference in measurements between examiners was less than 5 degrees. Technically, one must differentiate between the bony palate and vomer from a median sagittal view of the facial profile for optimal FMF measurements (Figures 44-14 and 44-15).

Fetal Iliac Angle

Since the late 1990s, several studies have correlated iliac angles—from a 2D axial view of the fetal pelvis—with the risk of Down syndrome.^{48,49, and 50} Bork et al⁵⁰ prospectively scanned 377 singleton fetuses, of which karyotypes were available for only 128 cases. The mean iliac angle for normal fetuses (68.2 ± 15.4 degrees) was significantly lower than observed in abnormal fetuses (98.5 ± 11.3 degrees). A receiver–operator curve for their high-risk population identified an optimal cutoff of 90 degrees for a detection rate of 90.9% (5.5% false-positive rate). By comparison, Shipp et al⁵¹ prospectively measured iliac angles for all fetuses undergoing second trimester amniocentesis over a 17-month period. Nineteen fetuses with Down syndrome and 1167 normal controls were scanned, with a mean iliac angle of 80.1 ± 19.7 degrees for abnormal cases as opposed to normal controls (63.1 ± 20.3 degrees). The iliac angle was at least 90 degrees in 36.8% (7 of 19 fetuses) with Down syndrome and in 12.8% (64 of 500 fetuses) with normal karyotype. Despite their results, it was concluded that the iliac angle alone was not useful in a high-risk population because of the high false-positive rate of 12.9%. French investigators also found a similar high false-positive rate of 20% for use of the iliac angle as a single marker for trisomy 21.⁵² Massez et al⁵³ examined the effect of fetal position on the iliac angle measurement using 2DUS in 695 fetuses during the mid-trimester. In euploid fetuses, the mean iliac wing angle was 83.7 degrees in decubitus and 68.7 degrees in the lateral position. In fetuses with trisomy 21, the mean angles were 104.9 and 102.5 degrees, respectively.

A relatively high false-positive rate of prior studies that utilized 2DUS may have been largely attributed to the complex structure of the fetal pelvis. To address this possibility, Lee et al⁵⁴ used 3DUS to standardize iliac angle measurements from multiplanar views of the fetal pelvis. Thirty-five normal fetuses and 16 fetuses with trisomy 21 were scanned during second trimester amniocentesis. The mean iliac angle for normal fetuses was 79 ± 5.5 degrees, which was significantly less than the abnormal fetuses (87.7 ± 4.9 degrees) (P < .001). Intraclass correlation analysis suggested that this technique was reproducible between examiners. For a false-positive rate of 5%, an axial iliac angle threshold of 87 degrees alone correctly classified 56% of fetuses with trisomy 21 in this high-risk group. A subsequent study described how skeletal rendering of the fetal pelvis can be used to reproducibly measure both axial and coronal iliac angles for predicting risk of Down syndrome. The multiple regression model had a sensitivity of 94.4% for a false-positive rate of 5% in the detection of fetuses with trisomy 21 from a selected high-risk population.⁵⁵ Further studies will be required to improve our understanding of the potential utility for iliac angle measurements for an unselected low-risk population during the second trimester of pregnancy.

Embryonic Volume

Falcon et al⁵⁶ used 2DUS and 3DUS to measure fetal trunk and head volume in 140 chromosomally abnormal fetuses at 11.0 to 13.9 weeks' menstrual age and compared them to 500 normal controls. In 72 fetuses with trisomy 21 and 14 fetuses with Turner syndrome, the crown-rump length (CRL) for gestation was similar, but the fetal trunk and head volume was about 10% to 15% lower. In fetuses with trisomy 18 (n = 29), trisomy13 (n = 14), and triploidy (n = 11), the deficit in volume was about 45%, as compared to CRL, which was lower by less than 15%. These results raised the possibility that fetal trunk and head volume measurements may have advantages over 2DUS for the identification of early fetal growth abnormalities. The same research group used similar techniques to show that early asymmetric growth restriction between the trunk and head was characteristic of triploidy, trisomy 13, and trisomy 18.⁵⁷ By comparison, the growth abnormalities equally affected the head and abdominal volumes in fetuses with trisomy 21 and Turner syndrome.

Several 3D volume analysis tools can be applied to the fetus. Although we are trained to mentally translate 2D images into 3D representations, this traditional approach is limited by the examiner's prior experience and ability to interpret this information. Furthermore, there are new volume analysis tools that permit us to visualize images in ways that are not possible using conventional methods. Despite these possible advantages, many diagnostic imaging cases can be simply evaluated from using only 2DUS. In some instances, 3DUS offers a complementary approach that may improve diagnostic confidence for the diagnostic impression that is initially based on conventional sonography.

We currently use 2DUS for the prenatal detection of most congenital anomalies, with a targeted application of 3DUS to answer specific questions that are raised from the initial diagnostic impression. As more applications for 3DUS are described against an emerging backdrop of technological improvements, the paradigm for how volume sonography is applied to obstetrical practice may also evolve.⁵⁸ For example, Benacerraf et al⁵⁹ have described how 3DUS improved the workflow of clinical practices by the efficient acquisition and review of volume data sets. Although this is not a focus of this chapter, several investigations have also proposed automating the analysis for data sets for the fetal heart.^{60,61,62, and 63} Others have reported remote sonographic diagnosis for telemedicine applications, using 3DUS in areas that are remote from the expert consultant.^{64,65,66, and 67} Volume sonography has also been used to assess fetal urine production in both normal fetuses^68,69 and after laser surgery for a case of twin-to-twin transfusion.⁷⁰

Many scientific articles have been published over the past decade as a result of the ultrasound manufacturing industry's successful efforts to commercialize 3DUS technology into clinical practice. This process has not only included improved image quality, smaller transducers, and faster computers, but the development of volume data analysis tools as well. Potential advantages and technical limitations of 3DUS will now be reviewed for selected obstetrical problems.

Face

The fetal face is important because it can provide diagnostic clues for the presence of isolated abnormalities and genetic syndromes. Although many patients immediately recognize facial features from the surface-rendered display, we believe that 3D multiplanar images are often the most helpful for medical diagnosis. However, the most appropriate selection of volume analysis tools critically depends on the question being asked. For example, one would use the maximum-intensity projection algorithm to visualize problems with bony structures such as the cranial sutures. Soft tissue clefts might be well visualized with surface rendering. The lips and hard palate could be systematically evaluated by using multiplanar images as well. This modality allows the use of a reference dot to improve understanding of complex anatomic relationships that are being investigated. Others might want to confirm a specific finding by using a parallel-slice display (eg, tomographic ultrasound imaging, multi-slice, i-Slice), or even thick-slice scanning of the fetal lips (eg, volume contrast imaging). Therefore, one must choose the most appropriate set of volume analysis tools on the basis of the diagnostic questions that are clinically relevant.

Many early investigators pointed out the benefits of 3DUS in evaluating the fetal face.^{71,72, and 73} Pretorius et al⁷⁴ described their preliminary experiences with visualizing the fetal face and lips using surface rendering and multiplanar views. In a subgroup of fetuses at less than 24 weeks' gestation, 3DUS confirmed a normal lip in 93% (58 of 63 cases) as compared with 76% (48 of 63 cases) using 2DUS. At that time, they noted that the 3D images of cleft lip were easier to understand for both the family and clinical colleagues. Merz et al⁷⁵ used multiplanar views of the face and found that the facial profile that was obtained by 2DUS represented the true midsagittal profile in only 69.6% of cases. In the remaining 30.4%, the profile view deviated from a true midsagittal section by up to 20 degrees in 1 or 2 planes. In this series, they found 20 of 25 facial anomalies that were demonstrated using both 2DUS and 3DUS. In the remaining 5 cases, 3DUS revealed additional anomalies that included 2 cases of narrow cleft lip as well as single examples of unilateral orbital hypoplasia, cranial ossification defect, and flat facial profile with decreased amniotic fluid volume. Such preliminary observations were quite extraordinary for the level of 3D technological development at that time. It was not until approximately 3 years later that the "electronic scalpel" image segmentation tool became commercially available on desktop computers.⁷⁶ This tool permits the examiner to selectively remove surrounding voxels that prevent the precise display of volume-rendered structures.

One approach initially applies 4DUS to confirm recognizable features of the fetal face. The rendered algorithms provided by the equipment vendors are often excellent and provide a rapid acquisition of the face (Figure 44-16). Movement of the mouth is also easily demonstrated. Acquisition is optimal when the fetal face is acquired from the sagittal or profile view by placing the region of interest box (ROI) directly over the face, with the rendering line just anterior to the nose (Figure 44-17). The rendering line may be straight or curved to optimize the image. Occasionally it is necessary to increase the threshold knob to take away unwanted echoes, particularly in heavier patients. If the face cannot be positioned in a sagittal orientation, a frontal view (coronal) can be obtained, resulting in a rendered image that displays the profile of the fetus.

Static 3DUS acquisitions of the fetus generally provide better resolution than 4DUS acquisitions and are used for diagnostic evaluation. The face can be acquired from any orientation, but sagittal and slightly oblique off sagittal are optimal for seeing the rendered face en face. Optimal planes of acquisition for various facial structures are summarized in Table 44-1.

Manipulation of the 3DUS volume should begin with attempting to get the face into a standard, symmetrical orientation. The cursor dot should be placed on a midline structure, preferably the nose or the region in between the orbits. The volumes should be rotated in all 3 planes until the orbits are symmetrical. It can then be evaluated by moving up and down in parallel slices in each plane.

The face can also be displayed using a multi-slice technique that is similar to magnetic resonance imaging or computed tomography where parallel slices at discrete intervals are varied to demonstrate the anatomy. In a series of 142 patients, McGahan et al⁷⁷ found that when starting at 3-mm intervals there was minimal manipulation needed to show in the axial plane the orbits, maxilla (primary palate), and mandible on one screen. Finally, Rotten and Levaillant⁷⁸ have nicely described how 2DUS and 3DUS can be systematically used to evaluate the fetal face.

Eyes

Although there are no published series of cases of abnormal orbits, case reports and images in review articles have been published.^{79,80, and 81} The orbits can be measured from volumes acquired of the face to identify hypotelorism (Figure 44-18), hypertelorism, and micro-opthalmia. We have found 3DUS very helpful when the orbits are absent or very small (Figure 44-19).

Periorbital masses such as dacrocystocele,⁸² frontal encephalocele, glioma, hemangioma, and teratoma may be difficult to evaluate with 2DUS, and 3DUS can be helpful.^{83,84, and 85} Several authors have reported that 3DUS was useful in evaluating these entities and for showing parents the 3D images for counseling.^{83,84, and 85}

Ears

Fetal ear abnormalities are often associated with aneuploidy (eg, trisomies 13, 18, 21) as well as genetic syndromes such as Treacher Collins syndrome, Fraser syndrome, CHARGE association, and VACTERL association.^86,87 The ears may be small, large, an abnormal shape, or in an abnormal position. Although 2DUS can be used to assess the fetal ears, 3DUS has been found to be extremely helpful (Figure 44-20). Shih et al⁸⁶ evaluated 18 fetuses with abnormal ears, and using 3DUS and found the ear shape, ridge pattern, and helix development as well as cranial location, axis, and orientation of the ear was better recognized on 3DUS compared to 2DUS. Case reports of abnormal ears seen using 3DUS in fetuses with Treacher Collins syndrome have also been reported.^88,89 Nomograms for ear length and width measurements obtained with 3DUS have been reported as a potential screening test for aneuploidy.⁸⁷

Metopic Sutures

The metopic suture lies in the midline of the face above the nasal bone and is the space where the frontal bones comes together (Figure 44-21). Abnormal development of the metopic suture has been associated with facial dysmorphism, fetal brain malformations, chromosomal defects, and genetic syndromes.⁹⁰

The use of 3DUS to evaluate cranial sutures and fontanelles was first reported in 1994 by Pretorius and Nelson.⁹¹ Visualization of normal sutures has been reported in 120 cases by Dikkeboom et al⁹² and in 120 patients by Faro et al.⁹³ In general, it is easier to visualize the sutures at earlier gestational ages. The metopic suture have also been evaluated in the first trimester.^94,95 Holoprosencephaly is associated with an accelerated development of frontal bones and premature closure of the metopic sutures.⁹⁴ Similar changes were not observed in fetuses with trisomy 21.⁹⁵ Faro et al⁹⁶ have also described the presence of a widened metopic suture in fetuses with Apert syndrome.

Chaoui et al⁹⁰ later described 4 patterns of abnormal metopic suture development. The first pattern involved delayed development with a V- or Y-shaped open suture in normal fetuses at 12 to 16 weeks. A second pattern was a U-shaped open suture. The third pattern was premature closure of the suture in normal fetuses after 32 weeks. The fourth pattern resulted from additional bone between the frontal bones in fetuses with holoprosencephaly and agenesis of the corpus callosum. The other 3 patterns were observed in fetuses with facial defects involving the orbits, nasal bones, lip, palate, and mandible.

Cleft Lip and Palate

Despite some geographic differences, oral cleft defects are among the most common congenital abnormalities with a prevalence of approximately 2.0 per 1000 births during the mid-trimester of pregnancy.⁹⁷ One large Norwegian study recently reported 101 fetuses or newborns with facial clefts in 49,314 deliveries. Twelve percent of the affected cases were associated with chromosomal abnormalities and 18% were documented with syndromes.⁹⁸ Cleft lip and palate were probably the main reasons that 3DUS was initially developed for the detection of fetal anomalies. Many papers have been written on the technique and benefits of 3DUS in evaluating the lip and palate. It assists in evaluating the presence, the extent, and the appearance for communication with the patient and her family. Subtle deformities can be precisely evaluated using a stationary volume, rather than a moving fetus. Chmait et al⁹⁹ showed that even clefts, thought to be isolated on 2DUS and 3DUS, were found to be associated with abnormalities at birth in 22% (8 of 37) of fetuses.

Volumes are acquired from static 3DUS volumes to evaluate the lip and palate. Axial acquisitions angled slightly upward toward the top of the mouth are optimal for evaluation of the primary and secondary palate. Rendered images of the face are helpful to demonstrate the cleft lip to the family (Figure 44-22). Multiplanar imaging can be used to evaluate the primary palate and lip. Many of the early articles only used the multiplanar reconstruction to evaluate for cleft lip and palate.^100,101 Johnson et al¹⁰² studied 28 fetuses with cleft lip with or without palate and found that 3DUS was able to identify the cleft palate more frequently (19 of 22) than 2DUS (9 of 22). They also found that management was changed using 3DUS in that some patients elected to terminate the pregnancy and others elected to carry the pregnancy when they had planned otherwise. In another study, Chmait et al¹⁰³ evaluated 53 fetuses with cleft lip with or without cleft palate and found that the diagnostic accuracy was improved for cleft lip to 100% (53 of 53) using 3DUS versus 91% (48 of 53) using 2DUS, and for cleft palate it was 89% (47 of 53) for 3DUS versus 57% (30 of 53) for 2DUS. Wang et al¹⁰⁴ also demonstrated how the use of a parallel-image-slice display format ("extended imaging") can also be used to evaluate fetal cleft lip and palate.

The rendered display can also be used to evaluate the secondary (hard and soft) palate. Campbell et al¹⁰⁵ first published the demonstration of the hard palate using a "reversed face" technique. He later wrote an informative editorial discussing issues related to evaluation of the hard palate with various rendering techniques.¹⁰⁶ Platt et al¹⁰⁷ described the "flipped face" technique to display the primary (alveolar ridge) and secondary (hard and soft) palate. This view emphasizes the palate with an upright face and viewing from an inferior direction, rather than superiorly (Figures 44-23 and 44-24). Faure et al¹⁰⁸ described a similar technique to display the primary and secondary palate in 100 low-risk fetuses that showed how 3D reconstruction of the fetal palate can be correlated to anatomic specimens. Pilu and Segata¹⁰⁹ studied 15 normal fetuses and 1 fetus with cleft lip and palate and showed that the secondary palate could be evaluated if the face was insonated at a 45-degree angle in the sagittal plane; in addition, they showed that the palate could be displayed on both axial and coronal planes (Figures 44-25 and 44-26). Faure et al¹¹⁰ also found that a 30-degree inclined axial plane appears to be useful for assessing the integrity of the fetal soft palate (Figure 44-27).

These examples serve to demonstrate the versatility of high-quality voxel-based images for providing several options for the analysis of fetal anomalies. It is important to remember that scanning artifacts may lead to misinterpretation of these images. For example, Nelson et al¹¹¹ have reported several artifacts related to both 2DUS artifacts being propagated through the volume and new artifacts related to 3D reconstruction, display, and scanning.

Jaw Abnormalities

An evaluation of the median facial profile can be performed using both multiplanar and rendered displays (Figure 44-28). The face should be rotated into a standard, anatomic orientation so that the face is symmetrical. Three-dimensional US has been shown to be useful in identifying abnormal profiles such as micrognathia (Figure 44-29), retrognathia, midface hypoplasia, and frontal bossing (Figure 44-30). Lee et al¹¹¹ evaluated 9 cases of micrognathia and found that an oblique plane can lead to misinterpretation of a normal chin to be abnormal, and that prominent cheeks (eg, in diabetic mothers) and retrognathia can lead to an overcall of micrognathia.

Evaluation of the fetal mandible can provide important diagnostic clues to the presence of genetic syndromes. By mid-2008, the Online Mendelian Inheritance in Man (OMIM) database cited 328 syndromes for micrognathia (small jaw) and 53 syndromes for retrognathia (posteriorly displaced jaw). From a clinical perspective, these jaw abnormalities can also be associated with obstructed airways and problems with feeding after birth. Toward this end, Rotten et al¹¹³ have described an objective approach for the assessment of abnormal jaw size and position using 3DUS.

Micrognathia was defined on the basis of a ratio between the mandibular width (MD) to the maxillary width (MX). Both parameters can be measured on an axial plane, caudal to the cranial base, at the level of the maxillary or mandibular tooth buds. From an axial plane, a line was extended posteriorly, 1 cm from the anterior border of the tooth buds. The MD/MX ratio was obtained from these 2 measurements (Figure 44-31).

Retrognathia was defined by an inferior facial angle (IFA) that resulted from the crossing of 2 lines (Figure 44-32):

1. A "reference line," orthogonal to the vertical part of the forehead, drawn at the level of the synostosis of the nasal bones.

2. A "profile line," joining the tip of the mentum and the anterior border of the more protrusive lip.

Their results demonstrated that micrognathia and retrognathia can be distinguished by the methods in this paper. The MD/MX ratio assesses mandible size, whereas the IFA evaluates mandible position. Although it is possible to obtain these measurements using 2DUS, the authors believed that 3DUS could be applied more easily for this analysis. Such improvements in our ability to more precisely diagnose congenital abnormalities may translate into improved therapeutic approaches and outcome.

Brain

Three-dimensional ultrasound images of the embryonic brain^6,9 have offered a fascinating insight into normal and abnormal development of the central nervous system. Similar to 2DUS, however, the fetal brain can be poorly visualized as a result of acoustic shadowing from the bony calvarium during the second and third trimester of pregnancy. A successful sonographic study of the brain often relies on the use of transvaginal sonography through a suitable acoustic window such as the anterior fontanelle.

Some of the earliest reports on the use of 3DUS for the diagnosis of fetal brain abnormalities included unilateral megalencephaly, hydrocephalus, anencephaly, holoprosencephaly, Dandy-Walker cyst, enlarged cisterna magna, periventricular leukomalacia, agenesis of the corpus callosum, and cerebellar fusion.^{114,115, and 116} Subsequent investigators have described 3DUS with varying degrees of successful visualization for fetal brain evaluation. One important technique for fetal brain evaluation was originally introduced by Timor-Tritsch et al¹¹⁷ as the "three-horn view." This view can be used to demonstrate the anterior, posterior, and inferior horns by either 2DUS or 3DUS scans. This 3D technique requires acquisition of a volume data set, orthogonal multiplanar display of brain structures, and rotation or tilting of the midcoronal section to the left or right. The examiner is able to navigate through the volume data for an improved understanding of these anatomic relationships for telemedicine, consultative, and teaching purposes.¹¹⁸ Another approach, the transfrontal view, uses the frontal cranial suture as an acoustic window to examine midline cerebral structures with abdominal 3DUS.¹¹⁹ This view was easily obtained in 89% of 124 healthy fetuses between 19 and 24 weeks' menstrual age. Anatomic findings were similar to a median sagittal scan through the anterior fontanelle. The transfrontal approach can also be used to evaluate the fetal facial profile. Correa et al¹²⁰ prospectively scanned 202 fetuses using abdominal 3DUS between 16 and 24 weeks' menstrual age. Acceptable cerebral multiplanar views were satisfactorily viewed by a sonologist who had expertise in neonatal cranial sonography in 92% of cases. The visualization rate for brain structures is summarized for the "3D multiplanar neuroscan" in Table 44-2. Pilu et al¹²¹ have also presented a very informative review that demonstrates how they use 3DUS to systematically evaluate the fetal brain.

As the imaging resolution and visualization software improved, other investigators examined the fetal brain as well. Roelfsema et al¹²² applied the Virtual Organ Computer-Aided Analysis (VOCAL) technique to document median brain volume that increased from 34 mL at 18 weeks to 316 mL at 34 weeks' menstrual age. This represents a nearly 10-fold increase during the second half of pregnancy. Viñals et al¹²³ later applied a thick-slice scanning technique (volume contrast imaging) to describe the normal appearance and dimensions of the fetal cerebellar vermis. Paladini and Volpe¹²⁴ found that posterior fossa abnormalities can usually be characterized using 3DUS on the basis of key findings that include upward displacement of the tentorium, counterclockwise rotation, and vermian hypoplasia of the cerebellum. One group has even described how 3DUS findings over time (ie, four-dimensional ultrasonography or 4DUS) can be used to assess neurobehaviorial movements of the fetus.¹²⁵ Roeflsema et al¹²⁶ described the normal development of the fetal skull base and found that these measurements were reproducible between examiners. In another study, serial volume measurements of the fetal cerebellum were prospectively acquired in 52 normal pregnant women between 20 and 32 weeks' menstrual age.¹²⁷ These volume measurements were taken every 2 weeks as a potential tool for fetal growth evaluation. All of these studies demonstrate the evolving applications of 3DUS for fetal brain assessment. In many cases, the diagnostic results from both 2DUS and 3DUS should be complementary to magnetic resonance studies of the fetal brain.¹²⁸

Spina Bifida

The prenatal diagnosis of spina bifida has steadily improved as a result of maternal serum α-fetoprotein screening and from the widespread use of ultrasonography. Corresponding advances in patient management have greatly reduced mortality, but have had minimal impact on long-term disability from neurological sequelae. This morbidity includes paraplegia, sensory deficits, spinal deformity, bowel dysfunction, and urinary incontinence.

Accurate characterization of spina bifida relies on sonographic recognition of disrupted ossification centers and/or overlying skin from transverse and coronal views of the fetal spine. Sonography predicts the clinical severity of open spina bifida because neurological symptoms correlate with the anatomic level of the defect. Kollias et al¹²⁹ reported that two-dimensional ultrasonography (2DUS) estimated the defect to within 1 vertebral segment in 79% of fetuses with spina bifida. However, 1 retrospective study of 171 consecutive cases of spina bifida found that only 29% of cases accurately identified the specific upper level of a spinal lesion using 2DUS.¹³⁰ Other investigators have proposed that 3DUS may be used to further characterize spina bifida.^{131,132,133, and 134}

Optimal views can be generated by manipulation of a virtual cutting plane through a volume reconstruction of the fetal spine. Lee et al¹³⁵ have proposed that 3DUS can be used as a semiquantitative technique for determining the anatomic level of this lesion. Multiplanar views are acquired using a volume probe from an axial sweep of the fetal spine (Figure 44-33). A coronal view of the lumbar and sacral spine is rendered with a maximum intensity projection (MIP) algorithm that primarily displays the bony spine. An electronic cutting plane is used to display orthogonal views of the volume-rendered spine, beginning at the spinal segment that is contiguous with the last fetal rib. As the examiner moves the cutting plane towards the sacral spine, simultaneous views of the axial spine and overlying skin line can be visualized. Other 3DUS techniques can be used to directly render the bony and soft tissue defects (Figures 44-34 and 44-35). Spinal defect levels obtained in this manner closely correlate with findings from both 2DUS and postnatal results. Although multiplanar views are generally more informative than rendered views for localizing these defects, their simultaneous use increases the likelihood that a spinal defect will be appropriately analyzed. This approach may improve the characterization of spina bifida by adding diagnostic information that is complementary to the initial assessment by 2DUS.

Skeletal Dysplasias

Volume sonography offers an important method that allows the examiner to visualize fetal skeletal structures on the basis of a MIP technique. This software algorithm displays only the most echogenic structures and it is possible to mix varying degrees of volume-rendered soft tissue for the final output display. Pretorius and Nelson were among the first to use 3DUS to visualize cranial sutures and fontanelles¹³⁶ as well as the thoracic skeleton.¹³⁷ The same group later demonstrated the value of stereoscopic imaging for these 3D reconstructions in order to better visualize fetal bony structures.¹³⁸

One of the earliest applications of 3DUS for a fetal skeletal abnormality was described in a fetus with platylospondylic lethal chondrodysplasia.¹³⁹ Yanagihara and Hata¹⁴⁰ subsequently reported early experiences with a specially developed abdominal 3D transducer for visualizing skeletal structures in 42 normal fetuses and in 3 anencephalic fetuses. Garjian et al¹⁴¹ reported a small series of fetuses with skeletal dysplasia using this technology. The series included campomelic dysplasia, thanatophoric dysplasia, osteogenesis imperfecta, arthrogryposis, and short-limbed dysplasia. Surface rendering, volume rotation, and multiplanar displays were especially helpful in 3 of 7 cases, when compared to 2DUS. Multiplanar imaging was helpful for displaying a true median facial profile, and volume sonography provided a more global assessment of the skeletal anatomy. The MIP technique made it possible to identify scapular hypoplasia in a fetus with campomelic dysplasia. The added depth perception cues and ability to rotate volume data both facilitated increased appreciation of positional limb anomalies as well as improved visualization of the spine. Another series from Cedars-Sinai Hospital further described diagnostic advantages of 3DUS over 2DUS for additional skeletal dysplasias for evaluating fetuses with facial dysmorphism, relative proportion of appendicular skeletal elements, as well as the hands and feet.¹⁴² Benoit¹⁴³ provided additional examples of the maximum-intensity mode for displaying normal skeletal structures, hemivertebrae, cranial sutures, spina bifida, hand digits, and cranial fontanelles (Figures 44-36,44-37,44-38, and 44-39).

Ruano et al¹⁴⁴ have reported the use of a novel technology, 3D helical computed tomography (CT), for the diagnosis of fetal skeletal anomalies between 27 and 36 weeks' menstrual age. There were 3 cases of achondroplasia, 2 cases of osteogenesis imperfecta type II, and 1 case of chondrodysplasia punctata. The correct diagnosis was made in 4 cases using 2DUS. Both 3DUS and 3D helical CT were used to make the correct diagnosis in all cases. However, 3D helical CT provided an advantage of imaging the entire fetus.

Tissue Perfusion Using 3D Vascular Flow Indices

3DUS and the VOCAL technique have been combined to estimate organ tissue perfusion using vascular flow indices (4D View, version 7.0, GE Healthcare, Milwaukee, WI, USA). Pairleitner et al¹⁴⁵ first introduced this approach for quantifying the degree of vascularized tissue in adnexal masses. First, the VOCAL technique is used to define a volume of interest that contains both gray-scale (nonvascular) and color voxels (vascular). The fundamental principle is based on the use of histograms to analyze the ratio of colored voxels to gray-scale voxels in the following manner.

1. VI = vascularization index (standardized range 0 to 100)

   

2. FI = flow index (standardized range 0 to 100)

   

3. VFI = vascularization flow index = (VI *FI)/100

   

Several investigators have used vascular flow indices to study gynecologic conditions. Pan et al¹⁴⁶ studied differences in ovarian stromal flow in patients with polycystic ovarian syndrome. The VI, FI, and VFI were significantly higher in affected women compared with women having normal ovaries. Raine-Fenning et al^147,148 also examined periodic changes in endometrial development and subendometrial vascularity during the normal menstrual cycle and found significantly reduced 3D vascular flow indices in women with unexplained subfertility. Ng et al¹⁴⁹ found significantly higher endometrial and subendometrial vascularity in pregnant patients with a live birth following in vitro fertilization using frozen thawed-embryo transfer cycles as compared to these women who experienced miscarriage.

Others have applied 3D vascular flow indices to the fetal liver,¹⁵⁰ brain,¹⁵¹ kidneys,¹⁵² and placenta.¹⁵³ All flow indices significantly increase over time with advancing pregnancy. Mercé et al¹⁵⁴ evaluated the reproducibility of placental vascular flow indices on 30 normal singleton pregnancies from 14 to 40 weeks' menstrual age. All 3D power Doppler vascular indices (VI, FI, and VFI) showed a correlation greater than 0.85, with a better intraobserver agreement for the flow indices (FI and VFI). They subsequently described increases of placental vascular flow indices over time.¹⁵⁵ Power Doppler settings included a PRF 600 Hz, wall filter 40 Hz, and fixed 35-degree field of view, although the overall gain settings were not reported. The FI, reflecting placental flow, increased in a linear and progressive manner as compared to VI (number of placental vessels) that increased up to the 30th week. The VI maintained a plateau up to the 37th week and decreased afterwards.

A few salient points are pertinent to this line of investigation. First, these noninvasive Doppler techniques do not measure true tissue perfusion because this technique would require knowledge regarding how fast blood volume moves through tissue. Second, the reproducibility of vascular flow indices is dependent on several technical factors such as system gain, pulse repetition frequency, angle of insonation, wall filter settings, and examiner experience. In this regard, the use of vascular flow indices may be more suited towards gynecological imaging, in which the proximity of the ovaries and endometrium to the transvaginal probe almost eliminates the beam path.^147,156 By contrast, the use of standardized ultrasound system settings may not satisfactorily compensate for different beam paths between obstetrical scans because of the signal attenuation that results from insonating intervening tissue at various depths.^157,158

To address some of these concerns, Rubin et al¹⁵⁹ have introduced the concept of fractional moving blood volume. In this elegant in vitro experiment, they used power Doppler ultrasonography to detect moving scatterers in a flow tube as a function of successive dilutions of a blood mimicking perfusate. Power Doppler ultrasonography has advantages over color Doppler methods because of its low noise variance, decreased electronic noise, greater dynamic range, and increased sensitivity to the detection of blood flow. The fractional moving blood volume technique normalizes power Doppler measurements to compensate for signal attenuation and insonation depth. These signal intensities can be compared to similar measurements from the center of a large blood vessel that is used to define the highest value for vascular amplitude scanning. These colored voxels are set to a value of 100%, and these normalized data compensate for the attenuation of the Doppler signal through tissue. Fractional moving blood volume has also been shown to correlate well with regional blood flow perfusion on the basis of radioactive microspheres in exteriorized fetal lambs that were exposed to experimentally induced asphyxia.¹⁶⁰ This technique also appears to be reproducible in the fetal lung during the third trimester of pregnancy.¹⁶¹ Furthermore, fractional moving blood volume has more recently been used to study regional cerebral blood perfusion in third trimester fetuses.¹⁶² Interobserver agreement showed a mean difference of −0.2 (SD 2.7) with 95% limits of agreement ranging from −5.6 to 5.2. Statistically significant differences in fractional moving blood volume were also found between cerebellar and complete sagittal planes of the frontal and basal ganglia regions. Future studies will need to address the role of fractional moving blood volume analysis for standardized 3DUS as well.

Abnormal fetal growth is associated with a significant proportion of newborn babies that develop postnatal complications. Approximately 8.3% of babies are born with low birth weight (<2500 g) in the United States.¹⁶³ These infants can experience a broad range of problems that include respiratory distress syndrome, intraventricular hemorrhage, patent ductus arteriosus, necrotizing entercolitis, and retinopathy of prematurity. Furthermore, the National Center of Vital Statistics also reports that 10% of all liveborn infants in this country weigh more than 4000 g. The most serious complication of fetal macrosomia is shoulder dystocia during delivery, although clavicular fractures and brachial plexus injuries from birth trauma also occur.^164,165 Contemporary obstetrical practice has relied heavily on a generalized assessment of fetal nutritional status that is based on either estimated weight or only a single abdominal circumference measurement. Fetal weight estimation models typically include the abdominal circumference with some combination of head or limb measurements. However, these fetal growth parameters do not emphasize the degree of soft tissue development. This issue is extremely important because there is mounting evidence of an association between low birth weight and health in later adult life.

The Barker hypothesis has stimulated much interest regarding the fetal origins of adult disease.¹⁶⁶ This line of investigation proposes that some fetuses will respond differently to an unfavorable intrauterine environment, and that malnourished fetuses will generally respond less favorably to these insults. Epidemiologic studies have found that low-birth-weight infants are at increased risk for decreased longevity as well as the onset of heart disease, diabetes, chronic renal disease, metabolic syndrome, and neurological sequelae in adult life.^{166,167,168, and 169} Although many of these studies have been based on low-birth-weight infants, obstetricians are now challenged with the task of separating fetuses that are small but otherwise normal from those that are truly malnourished. Early identification of malnourished fetuses could potentially be used to guide timely prenatal intervention, to minimize risk of adverse perinatal outcome, and possibly to reduce the development of health problems in later adult life. Nonetheless, is it possible that 3DUS could provide additional fetal size and growth parameters that will ultimately improve our ability to detect and monitor malnourished fetuses?

Several studies have described the early use of 3DUS for measuring fetal limb volume that includes soft tissue assessment.^{170,171,172,173,174, and 175} For example, Chang et al¹⁷⁰ investigated the relationship of fetal thigh volume to birth weight. The thigh volume was measured by manually tracing transverse slices of the thigh at 3-mm intervals, a process that usually took 10 to 15 minutes to complete for each limb. This information was used to derive the best one-parameter regression model in 100 women who delivered at term gestation. They prospectively compared their results in 50 additional Taiwanese subjects against previously published fetal weight estimation models.^{176,177, and 178} Their new model (birth weight = 1080.87350 + 22.447 × thigh volume) gave the most accurate and precise results when the data were expressed as mean percent differences (1.5 ± 5.6%) as compared to the methods of Warsof (−4.2 ± 9.4%), Hadlock (−4.8 ± 9.4%), and Thurnau (−19.2 ± 7.0%). Furthermore, they found no significant differences between the inter- and intra-examiner errors of mean differences among 20 of their subjects. Similar results with only fetal arm volume were prospectively obtained (0.35 ± 4.6%), although no reproducibility testing was performed between examiners.¹⁷² German investigators¹⁷³ subsequently reported new fetal weight estimation models that combined upper arm and thigh volume measurements with biparietal diameter and abdominal volume. The multiple parameter model prospectively yielded a mean percentage error of 1.9 ± 7.6%, as compared to results using the Hadlock model (0.8 ± 9.3%). No information was provided about the time necessary to complete these measurements and only intraobserver reproducibility results were reported.

Unfortunately, 2 major technical limitations have prevented the practical application of soft tissue measurements into obstetrical practice. First, limb volume measurements originally required a series of tedious manual traces around the axial limb from one end of the femur to the other. Second, soft tissue borders are often difficult to clearly visualize at either end of the long bone shaft due to acoustic shadowing. This inherent limitation of sonographic imaging can make manual tracing of soft tissue borders difficult from overlying dense tissue. Lee and colleagues¹⁷⁹ addressed these technical concerns by introducing the concept of fractional limb volume (Figure 44-40). A commercially available software product (4D View, version 9.0, GE Healthcare, Milwaukee, WI, USA) can be used to measure fractional limb volume that is based on an analysis of a midsection of arm or thigh from a volume dataset (Figure 44-41). The examiner is able to more rapidly measure soft tissue volume in the center of the limb, where the manual tracings around the soft tissue borders are less likely to be obscured by acoustic shadowing. Manual fractional limb volume measurements take approximately 2 minutes to complete and are reproducible between observers throughout a broad range of pregnancy. The random error of fetal weight estimation appears to be improved by adding fractional limb volume measurements to conventional 2D biometry with best results in the range of 5.6 to 6.6%.^{180,181,182, and 183} Even if the precision of these weight predictions is similar to what is already provided by conventional 2D models, this new approach now adds a soft tissue developmental parameter that may allow better detection of malnourished infants (Figure 44-42). However, larger prospective validation studies will be required to confirm the clinical impact of these observations. Fractional limb volume has also been successfully applied to individualized fetal growth assessment using the Rossavik fetal growth model (Figures 44-42 and 44-43).^184,185 Normal values for fractional limb volume and the reproducibility of these measurements have also been reported during pregnancy.¹⁸⁶

Another study used air displacement plethysmography for newborn infant body composition to determine which prenatal growth parameters had the greatest correlation to percent body fat (%BF).¹⁸⁷ Fractional thigh volume (TVol) was the most significant predictor of %BF and this parameter explained 46.1% of the variability in %BF. Actual birth weight similarly explained 44.7% of the variation in %BF. By comparison, abdominal circumference and estimated fetal weight accounted for only 24.8% and 30.4% of the variance in %BF, respectively. Skeletal growth parameters, such as femoral diaphysis length (FDL; 14.2%), head circumference (HC; 7.9%), and BPD (4.0%), contributed the least towards explaining the variance in %BF (Figure 44-44). Moyer and colleagues have subsequently confirmed the limitations of conventional 2D ultrasound measures of fetal growth for prediction newborn adiposity that is measured using air displacement plethysmography.¹⁸⁴ If a major goal of prenatal care is to detect malnourished fetuses, the rationale for using single parameters such as abdominal circumference (AC), or even estimated fetal weight (EFW), to detect and monitor fetal growth abnormalities may require critical re-examination. However, fractional thigh volume still accounts for no more than one-half of the variance in neonatal %BF. Other novel fetal soft tissue parameters must be identified and validated to improve our understanding of soft tissue development and, possibly, the subsequent risk for disease in later life. Supplemental information for fetal assessment using fractional limb volume can be found by visiting http://www.obsono.org.

Safety Issues

The 2 main bioeffects of ultrasound exposure to fetuses are based on the possibility of thermal effects and mechanical cavitation.^188,189 Many professional organizations encourage the prudent use of diagnostic sonography for medically indicated scans by trained health care providers during pregnancy.¹⁹⁰ Specific emphasis is placed on the examiner for using acoustic power settings that are as low as reasonably achievable for the acquisition of satisfactory diagnostic images (ALARA principle) with the corresponding use of appropriate output display standards.¹⁹¹ A recent review of this topic suggests that exposure to diagnostic ultrasound does not pose a significant health risk to either the expectant mother or her fetus.¹⁹² Sheiner et al¹⁹³ have performed a prospective, longitudinal study of 3 commercially available 3D ultrasound systems to examine the mechanical and thermal indices during pregnancy. Mean thermal indices from volume sonography were comparable to corresponding values resulting from conventional 2D ultrasound scans. However, the mechanical indices for volume acquisitions were significantly lower than observed using 2D B-mode ultrasonography.

Parental Bonding

It has been known for some time the 2DUS imaging increases bonding between parents and fetuses.^{194,195,196, and 197} As practitioners began using 3DUS, it appeared that there was an even more obvious recognition of the fetus by parents, particularly images of the fetal face, and it appeared that there might be increased bonding between parents and their fetus. Many physicians hoped that an increase in bonding might lead to a decrease in abusive behaviors such as child abuse, spousal abuse, and maternal smoking and use of drugs and alcohol, however no significant data have yet been published.

Several studies have now been published looking at bonding between parents and the fetus after observing 3DUS images of their fetus. Certain subsets of patients appear to benefit more from viewing the images, and these include patients having experienced prior anomalies or fetal demise, surrogate parents, hospice patients carrying fatal anomalies, patients with infertility issues, and patients with a family history of anomalies.¹⁹⁷ In a study of 100 patients, Ji et al¹⁹² looked at 50 patients having undergone 2D ultrasound and 50 patients having undergone 3D ultrasound. They found that the patients who had 3D ultrasound shared their photographic images with more people (median 27.5) than those who had 2D ultrasound (median 11.0). In another study by Pretorius et al,¹⁹⁸ 189 patients filled out, both before and after 3DUS imaging, a validated questionnaire modified for pregnant parents and found that bonding had increased.

Drawings of fetuses made by parents have been evaluated by 2 different teams of investigators and appeared to increase bonding after undergoing 3DUS.^199,200 It was possible to predict which pictures were drawn before and after the 3DUS examination primarily based on uterine context and positioning of the extremities.²⁰⁰ Parents appear to have a more realistic view of their fetus after seeing three-dimensional ultrasound images.

The data regarding bonding are, however, difficult to assess. Rustico et al²⁰¹ found that facial expressions and hand-to-mouth movements were twice as likely after 4DUS in 48 patients (44%) compared to 2DUS in 52 patients (27%). The 4DUS group had more women with positive quality, intensity, and global attachment, although this was not statistically significant. In addition, fetal structures and movements visualized were similar between the 2 groups, as the face was visualized in 83% after 2DUS and in 81% after 4DUS. No difference was found in the percentage of women who felt that the scan had changed their perception of the fetus for the better.

Medicolegal Issues

A common question regarding medical legal use of 3DUS is related to whether the volumes should be saved. It is clear that much information is within each data volume, and the examination of all data is not feasible for all possible planes. Diagnostic images that are derived from volume data should be recorded as single images and saved. It is not mandatory that volume data sets be saved unless there is a need to archive them for research, consultative, or teaching purposes. This strategy is similar to how MRI and computed tomography data are archived and stored in clinical practice.

Ultrasound for Entertainment

Patients and medical care providers have various opinions regarding whether parents should be allowed to have 3DUS examinations for bonding alone. The question is whether ultrasound examinations are entertainment or reassurance. Three-dimensional US studies are now being formed in commercial venues as well as pregnancy crisis centers after minimal training, often only 2 weeks. The medical community is concerned about biohazards of diagnostic ultrasound, including the understanding of power output, risk of Doppler to the early embryo, length of time and frequency of scanning, as well as misdiagnosis if performed by nonskilled personnel outside the medical environment.

Both the Food and Drug Administration and the American Institute of Ultrasound in Medicine (AIUM) have guidelines recommending that diagnostic ultrasound only be used when there is a medical indication. The AIUM strongly discourages the nonmedical use of ultrasound for psychosocial or entertainment purposes. Physicians have varied opinions regarding the use of ultrasound for reassurance. Dr Peter Doubilet²⁰² suggested that to ban or condemn the use of ultrasound for entertainment is inconsistent with our acceptance of many other goods and services that provide enjoyment while carrying a small potential risk. Good examples of the computer, microscope, and camera were described that were all initially single-use devices and only later were used for many different purposes. This editorial reminds us that we do not forbid airline travel for pregnant women even though there is a cosmic ray–related risk. Noting that our societal approach is to warn mothers rather than to dictate actions, he also states that businesses make it clear that entertainment ultrasound is not a diagnostic medical sonogram.

Future developments for volume sonography will include faster imaging systems with improved detail resolution in both the original plane of volume acquisition as well as reconstructed planes. Ultrasound volume probes will become smaller and may eventually rely on the use of matrix array probes for real-time imaging, particularly of the fetal heart. New software applications will allow us to analyze and quantify volume data in ways that are currently not possible, and the efficiency of clinical workflow will also be improved with automated algorithms. It may eventually become practical to interactively explore four-dimensional volume data for the moving fetus within a virtual environment. This type of "virtual ultrasound simulation" would allow improved depth perception and allow joint discussion with colleagues for telemedicine applications, research, and training.^{203,204, and 205}

Volume sonography is an important complementary diagnostic technique to conventional 2D imaging.²⁰⁶ In this chapter, we have presented a perspective of the current medical literature with our extensive experiences in this area. Remember that 3DUS is not just one method, but it is based on a combination of several image visualization tools that must be customized to answer the pertinent clinical questions being asked. This technology can be used to improve diagnostic confidence as part of the problem-solving process. Three-dimensional ultrasonography is still in its infancy as software tools become more sophisticated and as we begin to apply these versatile capabilities for a wide variety of clinical problems. Health care providers should be aware of these emerging capabilities and their potential applications.

Highlighted References