Objectives

• Understand the principles of 3D ultrasound

• Discuss the principal applications of 3D ultrasound in obstetrics

Introduction

Two-dimensional sonography is the traditional way we have been using ultrasound in all areas of medicine to display normal and abnormal anatomy. Over these 30 years, we have witnessed an incredible evolution of sonographic imaging from A-mode all the way to exquisitely high-resolution grey- scale displays using highly sophisticated computer technology. More recently, three-dimensional (3D) ultrasound has emerged, allowing the acquisition of a volume and the display of any plane within that volume regardless of which orientation was first used to obtain the image. Surface rendering of the acquired volume is also possible with a myriad of different post-processing display techniques. Those who think that we have been using 3D sonography for only a short time are sorely mistaken. Thus far the most versatile 3D reconstruction techniques are inherent in the brains of those trained in standard two-dimensional imaging. It is the ability to turn a two-dimensional set of cross-sectional images into a three-dimensional understanding of anatomy which our brain has been performing for decades and which really represents the essence of cross sectional imaging in medicine. Only recently however have we been able to display digitally the actual image, which we have been conjuring in our minds for years. It is now possible to see how accurate this mental reconstruction of 2D images has been over the years since we can now display the actual 3D reconstruction on a monitor rather than in one's mind's eye. (Fig 1) The ability to display a 3D image has been one of the most powerful recent advances in sonography particularly in the field of obstetrics and gynecology. This paper will focus on 3D imaging in obstetrics and demonstrate the role of this new technology in the field of fetal imaging.

Click on image to view full size

Figure 1. Surface rendering of third trimester fetal face.

Three-dimensional imaging is basically two-dimensional imaging acquired as a sweep or volume and then displayed in various ways. The sweep acquisition can be done free hand or using a specialized transducer, which sweeps a volume mechanically and allows the processing of these volumes in a standard manner. The volume is then digitally stored and can be displayed either as a multi-planar image showing three orthogonal planes or as a surface rendered image. The three perpendicular planes display the X, Y and Z-axes with the understanding that the Z plane is one that cannot be acquired directly. (fig. 2) In fact by reorienting the initial acquisition plane after the volume has been acquired, the other two planes will follow, allowing us to view a standard two-dimensional cross sectional image in any plane within the volume. This method enables a sonologist to "rescan" a patient by reviewing the saved volume in any two-dimensional plane, even if different from the original scan plane. This is particularly useful if a fetus being imaged is not in an ideal position and the acquired volume can be manipulated to display the image in a non-scanning reconstructed plane. The image of a fetal profile, for example, is necessary to image the chin, and may not be obtainable on a fetus that is not positioned properly. The volume of the fetal face can be displayed in any desired orientation, including a sagittal view, to optimize the fetal profile or any other area if interest.

Click on image to view full size

Figure 2. Multiplanar reconstruction of a fetal face in the 3rd trimester, showing dacrocystoceles. The image was reoriented so that the views of the fetal face are standard.

We can then further manipulate these multi-planar displays using different post processing techniques. We can also navigate through the volume using a single point of reference in all three planes viewing its relative location as the planes move. There are different resolutions for acquiring the volumes, depending on how fast or how slowly the sweep is done. Clearly the fastest acquisitions have the lowest spatial resolution and vice versa. One can also vary the size of the area of interest by changing the angle of acquisition as well as the angle sweep or fan. An advantage to the automated sweep is the ability to use the reconstructed planes for volume measurements since the acquisition has been standardized.

Surface rendering is a process by which the surface of an area of interest can be displayed rather than individual cross-sectional slices. This image can be rotated in various directions to demonstrate the surface of the entire object. In addition, there are different post-processing modes available to enhance diagnostic capabilities, which can produce a smooth surface, a transparent surface, a bone type window etc. Many combinations of these post-processing possibilities exist to optimize the image. The images resulting from surface rendering are those that patients most appreciate when viewing their fetus in “three dimensions”. To further clarify the rendered image, one can use the electronic scalpel to eliminate superfluous areas and thus accentuate the areas of interest. The value of the electronic scalpel for the fetal face has been demonstrated by Merz et al. who demonstrated diagnostic improvement as well as superior image quality in two-thirds to three-fourths of their cases. (1) (fig 3)

Click on image to view full size

Figure 3. Surface rendering of third trimester fetal face, partly obscured by the placenta.

There are many useful applications for three-dimensional volume acquisition of two-dimensional ultrasound. The first includes networking and the ability to send packets of information from one site to another. The information obtained with three-dimensional volume acquisition is far superior to simply a video or cine-loop of two-dimensional information. Several studies have demonstrated the benefit of using three-dimensional volume sets to send to a remote location via electronic networks to a specialist, who will review it and render an interpretation. The specialist can then reorient the volume even if the imaging was not done in an ideal plane or if the fetus was not in a desired position. This enables sites distant from a central location to optimize their backup capabilities using remote “expert” consultation.

There is a definitely a learning curve to the ability to obtain a good volume set and the training of the sonographer or the physician obtaining the volumes as well as those reading it must be different than the training in standard 2D imaging. Training must include standard acquisitions of volumes that will display with a minimum of artifact. Sonographers must recognize inadequate volumes that cannot be used due to motion or other artifacts. The training of physicians reviewing the volumes must also include learning how to evaluate anatomy in orientations different from the original acquisition plane. A standardized protocol must also be in place so that the volumes are not viewed haphazardly but in a standard fashion and in multiple planes. (2, 3)

Further research has been done by Nelson et al. evaluating the feasibility of performing a “virtual patient examination” using three-dimensional ultrasound acquired in one location and sent to remote locations to be read. They demonstrated that overall, 3D ultrasound could be used with diagnostic quality results comparable to standard two-dimensional ultrasound, although the reconstructed 3D image quality itself was lower than the directly acquired two-dimensional image. There were also differences among reviewer’s interpretations thus emphasizing the need for a standardization of acquisition and reviewing protocols for both the sonographers and the physicians. Three-dimensional virtual exam is particularly well suited for echocardiography, where the volumes can be sent via the Internet to a tertiary fetal cardiology center to reevaluate a cardiac data set. Studies showed that a three-dimensional virtual examination of the fetal heart is possible although there are still limitations that need to be worked out. Certainly the main cardiac connections can be viewed and reconstructed in different ways. This may be helpful to patients being scanned in remote locations where questions about cardiac anatomy may occur. (4, 5)

Three-dimensional ultrasound has been touted as the novel sonographic method in fetal ultrasound that can enhance diagnostic capability over standard 2D ultrasound. Critics of the technique however feel that three-dimensional ultrasound has been over rated and that sonologists’ or sonographers’ training enables them to perform 3D reconstructions in their mind’s eye, with similar results as those described with actual 3D displays. In the next few paragraphs, we will evaluate several organ systems and present the data in the literature that suggests where 3D ultrasound can be used as an adjunct to the standard 2D imaging.

In general, there are several studies that suggest that three-dimensional ultrasound is advantageous in demonstrating fetal defects. Merz et al. studied 204 patients and found 3D reconstruction was helpful and in 62% of patients. They also showed that 36% of the time 3D imaging was disadvantageous due to movement artifacts and technical problems. Dyson et al. reported on 63 patients with 103 anomalies and found that 3D offered diagnostic advantages in about one half of the cases but only affected management in one patient. They found that 3D was an adjunctive tool to 2D providing a more comprehensive image although only to be used as a targeted study to compliment 2D ultrasound. (6, 7)

The fetal brain is one of the areas where 3D ultrasound has been most helpful. Doctors Monteagudo and Timor have demonstrated that 3D ultrasound with reconstruction of the third non-scanning plane is crucial in demonstrating planes of section not previously visible sonographically; for example the area of the corpus callosum is not readily imaginable in standard imaging planes. (8, 9) Other practitioners have found that imaging of the brain particularly in the first trimester is advantageous using three-dimensional “neurosonography” showing normal and abnormal anatomy in different orientations. (10) Not only is imaging of the brain important using 3D ultrasound due to the capability of displaying anatomy in novel planes, but also, evaluating the sutures in fetuses where there is a question of craniosynostosis is made simple using volume imaging. Surface rendering of the skull is instrumental in evaluating the status of cranial sutures and determining whether or not an abnormal head shape is secondary to craniosynostosis. (11, 12) Three-dimensional surface rendering is also particularly helpful in the evaluation of other surfaces of the fetal head and face, such as the fetal ears, which are not normally focused upon using standard 2D cross sectional imaging. (fig 4)

Click on image to view full size

Figure 4. Surface rendering of second trimester fetal face, showing the fetal ear and suture.

Surface imaging of the fetal face using 3D sonography has probably been the most notable area of study, not only in the ultrasound literature but also in the lay press where patients have been able to see the face of their unborn baby more clearly than ever before. (13,14)(fig 5) Although the trained sonographer and sonologists are able to reconstruct the fetal facial anatomy and determine whether or not there is a facial cleft using standard 2D cross sectional imaging, patients and/or referring physicians such as plastic surgeons benefit from the 3D display surface rendering which shows the face in a way they are used to seeing it in real life. The spacial resolution of the reconstructed image is obviously not as good as the original acquisition scan plane for any given machine and therefore in many cases the specific diagnosis of a facial cleft is usually made using standard 2D long before the 3D application is turned on. Using 3D surface imaging however enables the plastic surgeons to understand the extent of the lesion and to counsel their patients who can also visualize the abnormality. Although surface rendering of the face has received the most press, the three orthogonal planes (multiplanar reconstruction) of the face are also extremely important. They enable the practitioner to reorient the fetal face in a standard fashion even with a fetus in unusual positions. This then permits the practitioner to view a perfect fetal profile plane as well as a colonal plane regardless of the actual fetal position at the original volume acquisition. (fig 6) In many cases a perfect sagital view of the profile using two-dimensional imaging is not possible leading to the erroneous suggestion of micrognathia or other facial abnormalities. Being able to manipulate the volume such that the profile is oriented correctly can then facilitate the evaluation of the jaw position with respect to the rest of the face. Merz et al. studied 618 patients between 9 and 37 weeks using both a transvaginal and transabdominal approach using the three orthogonal planes. Only in 69% of the cases was a true midsagital profile obtained using two-dimensional scanning only. The three-dimensional reconstructed views allowed a viewing of the perfect midsagital profile in all cases. There were a total of 25 facial anomalies 20 of which were clearly detected both in 2 and 3D whereas in 5 cases there are additional features identified using only three-dimensional scanning. (15)

Click on image to view full size

Figure 5. Surface rendering of third trimester fetal face.	Figure 6. Multiplanar reconstruction of a fetal face in the 3rd trimester, also showing surface rendering image.

Everyone agrees that 3D sonography of the face is not a screening technique but an adjunct to a good 2D scans. To date, it is not yet possible to detect dysmorphologic features using 3D surface rendering although it is hoped that as the techniques improve, the dysmorphologic fetal facies will be detectable. (13-15)

It is the evaluation of the fetal cleft lip and palate that has gotten the most attention in the literature. (figs 7-9) Aro et al studied the ability of three-dimensional ultrasound to visualize fetal tooth germs, which are abnormal in fetuses with cleft palate and oligodontia. They showed that 3D imaging, using multiplanar reconstruction, was superior to conventional sonography for identifying the fetal tooth germs. The tooth buds were delectable in 80% of fetuses using 3D versus 56% of the time when using standard two-dimensional sonography. (16) Lee et al also demonstrated that interactive views of standardized 3D multi-planar images allowed the practitioner to visualize the fetal lip and palate and in particular the alveolar ridge and tooth buds. (17) It is possible to evaluate the presence and degree of pre-maxillary protrusion for fetuses with facial clefting. From a sonologist’s point of view the multi-planar images are probably more important than is the surface rendering for evaluating the extent of facial clefting, whereas the surface rendering is more important for the plastic surgeon and the patient to view the lesion. (17, 18)

Click on image to view full size.
Figure 7. Surface rendering of second trimester fetal face, showing the normal fetal mouth and lips.
Figure 8. Surface rendering views of third trimester fetus with unilateral cleft lip and palate.
Figure 9. Surface rendering of second trimester fetal face showing a subtle unilateral incomplete cleft lip.

Three-dimensional imaging of the normal and abnormal extremities has been studied extensively using both multi-planar reconstruction and surface rendering. (19)(Fig 10) Several studies have described fetal skeletal dysplasias viewed with 3D ultrasound, where practitioners found that 3D provided additional information on affected fetuses as compared to 2D. (fig 11-13) On the other hand, other investigators have not been as enthusiastic about the role of 3D for fetuses with these malformations, thus requiring further investigation to come to a consensus. Opinions that 3D provides diagnostic details not available using two-dimensional ultrasound remains somewhat anecdotal and restricted to small case series. (20-22)

Click on image to view full size.
Figure 10. Surface rendering of third trimester fetal arm.	Figure 11. Surface rendering of second trimester fetal foot, showing a clubfoot.
Figure 12. Thirteen week fetus with trisomy 13. Note the facial cleft (a) and the polydactyly (b).
Figure 13. Surface rendering of second trimester fetus with osteogenesis imperfecta. Note the short limbs with unusual angulations consistent with fractures.

Evaluation of the spine using 3D has been focused on determining the volume of the spine (see below) as well as the level of a spina bifida lesion previously detected using two-dimensional standard scanning. Determining the level of the spina bifida is of great importance in predicting outcome accurately. Precise information is crucial for the parents of the unborn fetus and the neurosurgeons who are counseling them. It is clear from several reports that although 3D made additional diagnostic information possible, the technique was not necessary for definitive diagnosis of spina bifida. (Fig 14,15) Three-dimensional ultrasounds however did display the level of the defect more accurately than with conventional scanning. Scoliosis resulting from vertebral body anomalies was also recognized more easily on a single three-dimensional rendered image whereas multiple standard 2D images were needed to make the same diagnosis. (23, 24) Click on image to view full size.

Figure 14. Multiplanar reconstruction of a fetus with a neural tube defect involving the lower spine. Note the defect on the surface rendering image (lower right).
Figure 15. Same neural tube defect as pictured in figure 12, showing that the image obtained directly in 2D (a) has better resolution than the image of the same area reconstructed from the volume (b).

Evaluation of the fetal heart is an intense area of research interest leading to an initial feasibility study done by Sklansky et al to determine whether volume clips could be stored off-line and sent to remote locations for viewing. Data could be displayed in multiple planes and cardiac motion could be slowly stopped as necessary to view the anatomy more easily. Most of the structures as well as the cardiac motion can be viewed appropriately and in real time. (25) Evaluation of the heart in 3D however can be tricky and requires Doppler gated capabilities. There is still much work to be done to bring 3D imaging of the heart up to a standard currently possible using 2D operated by trained personnel. (26)

Other areas of investigation include fetal genital scanning where the fetal perineum can be visualized, although to date, the performance of 3D technology is reportedly not as good in evaluating fetal genitals than is 2D imaging alone. Perhaps as technology improves, more work will be done in this area. (27)

The anterior abdominal wall can also be well demonstrated using surface rendering 3D although there is no data indicating that this can not be done using standard two-dimensional ultrasound alone. (Fig 16) It is the belief by some that 3D sonography of the anterior abdominal wall, while not necessary to make any specific diagnosis, can be a useful adjunct for more efficient counseling and postnatal planning by consulting pediatric surgeons. (28)

Click on image to view full size

Figure 16. Surface rendering of second trimester fetus with omphalocele.

Three-dimensional scanning of the early embryo in the first trimester is an area of intense research. According to several authors, 3D ultrasound has played an important role in “sonoembryology”. (29-32) The benefits of 3D imaging of the fetal brain have been mentioned earlier, particularly in the study of early brain development. (31) Multi-planar reconstruction of the non-scanning (Z) plane is particularly useful in evaluating the nuchal translucency measurement of fetuses 11-14 weeks, who are not oriented in a perfect sagittal position. It is clear that as many as half of fetuses presenting for measurement of nuchal translucency are not positioned such that an accurate nuchal translucency measurement is possible using a standard plane of scanning. Acquisition of the volume is quick and permits the display of the non-scanning plane for measurement. This is by far the most important function of first trimester three-dimensional scanning in my opinion.

Another important clinical application of 3D is volume measurements calculations based on 3D volume acquisition. Brummer et al. demonstrated the accuracy of volume based three-dimensional sonography measurement on follicle aspiration performed using transvaginal needle guided technique. (33) Other practitioners have also shown that three-dimensional sonographic methods provide accurate volume measurements of both regular and irregular objects thus providing an improvement, both in accuracy and examination time, over estimations using multiple 2D images (34, 35) Specific area of study where volume measurements are clinically applicable and currently under study, include serial fetal lung volume measurements for the prenatal detection of pulmonary hypoplasia. (36-38) The volume measurements of the fetal thoraco-lumbar spine as well as kidneys, liver and heart have been established, and fetal liver volumes in normally grown fetuses have been compared to those small-for-gestational-age. (39-44) Fetal brain volumes have been calculated using 3D scans with excellent intra and inter observer variability correlating well with other standard biometry at different gestational ages. (45) Three dimensional volume measurements may improve our ability to establish fetal weight estimates as well as growth parameters and improve the accuracy with which we can predict small-for-gestational-age infants. Fractional limb volume has also been investigated as an additional parameter to further improve accuracy in predicting birth weight, although this remains to be seen in larger scale studies. (46-48) Other attempts using 3D to estimate fetal weight have shown superior accuracy due to the inclusion of soft tissue volume measurements, although further work is needed to validate this. (49) Volume measurements of the fetal placenta using 3D sonography have not been particularly helpful to date in predicting small-for-gestational-age infants. (50)

Three-dimensional Doppler is an area of intense research using both power Doppler and traditional color flow. One particular area of interest has been the detection of abnormal placental adherence (accreta, inccreta, or perccreta) with or without bladder involvement. Three-dimensional color Doppler has allowed the investigators to visualize all three orthogonal planes of the placental myometral unit, a technique which appears to be complimentary to standard 2D sonography in clarifying abnormal neovascularization in patients with placenta accreta. (51, 52) Power Doppler in 3D is currently being studied as a way to visualize the fetal vascular system for prenatal diagnosis of anomalies, although more information is required to determine the utility of these techniques above and beyond the 2D Doppler. (53). Real-time capabilities in the use of Doppler will be necessary in 3D applications with similar frame-rates as those available today in 2D Doppler examinations.

Researchers have attempted to use real-time 3D (otherwise known as 4D) in the assessment of fetal behavior during pregnancy. Although the number of frames per second is still less than required for a smooth real time image, there is enough information using continuous three-dimensional sonographic images to display fetal activity. (Fig 17) It is unclear however how much more information is available using 3D than 2D realtime, since fetal movement is readily visible in 2D and has been for many years. It may however be possible to image more of the fetal body at once using real-time 3D surface rendering than using the single slice standard 2D imaging. Multi-planar displays in realtime (3D) may also be advantageous to visualize movement in a non-scanning (or Z) plane. It remains to be seen whether realtime 3D (or 4D) sonography will play a role in the evaluation of fetal well being during gestation. (54)

Figure 17. Two frames from a 4D surface rendered image of a fetus moving its arm.

Technical aspects of imaging must be considered as this new technology is learned by practitioners in our field. Not only are there artifacts inherent to 2D imaging present in 3D ultrasound but additional artifacts specific to volume imaging have also emerged. Such acoustic artifacts as dropout and shadowing which are well known to the ultrasound community are present in 3D imaging although more difficult to recognize due to different and unfamiliar displays. Color and power Doppler artifacts relating to gain and flash may also be confusing in rendered images. Three dimensional volume sets are hampered by fetal movement, cardiac motion, as well as movement of adjacent structures. Acoustic shadowing and other artifacts look very different when displayed in 3D volumes and may be more difficult to recognize than on standard 2D due to lack of specific training of personnel. These artifacts may produce apparent defects such as limb abnormalities or facial clefts where they are not present. Acquiring data from multiple orientations may avoid artifacts of this type. (55-57)

What can we glean from this large body of information, which has emerged as 3D ultrasound and has taken hold in our community? It is clear and must be emphasized that there has not been any randomized control trial of 3D versus 2D imaging and the published information reporting the benefits of 3D are at best speculative, anecdotal and consist of case series rather than well controlled studies. It is also apparent that evaluating a fetus from a point of view other than the scanning plane is greatly beneficial and has many applications, although fetuses do move around in the womb or can be reoriented to demonstrate other scan planes if one has the time to wait and induce fetal movement. It is still obvious to me that standard 2D imaging done on a top quality machine has better spatial resolution than 3D reconstruction of the same image on the same equipment, whether it be multi-planar or surface rendering. (Fig 15) Three-dimensional imaging is also time consuming for the practitioner, who is manipulating the image volume set long after the patient is gone; this kind of interaction with the volume set has a definite learning curve and may or may not provide improvements in any given case. The beautiful 3D images of the fetal face shown in the lay press can only be obtained in less than 20% of fetuses and problems such as shadowing by fetal limbs, oligohydramnios or positioning of the placenta can make imaging of the face in three-dimensions often difficult and frustrating. (58)

Why then should we adopt 3D imaging? Three-dimensional volume acquisition and displays represents an important milestone in the history of ultrasound where it is now possible to view a reconstructed image of a region which can not be imaged directly in the standard horizontal or vertical planes. Although it may be ultimately possible, particularly in the moving fetus, to obtain different planes of section, it may be enormously time consuming and occasionally impossible to reposition the fetus. Reconstruction of the third orthogonal plane enables the practitioner to evaluate any plane of section, regardless of fetal orientation and permit easy measurements not only of distance but volume. Volume measurements may become important in the evaluation of fetal growth, and the development of many organs, since standard 2D scanning has not permitted us to determine accurate volumes in the fetus. Additionally, surface rendering is very helpful in areas where 2D has not focused such as fetal cranial sutures, fetal ears as well as unusual fetal anatomy. (Fig 18) Evaluation of the lip and palate in fetuses with facial clefts is particularly helpful for our colleagues untrained in ultrasound (referring physicians and plastic surgeons) as well as the patients who need to understand the lesion and the treatment that will take place after delivery. Networking and sending acquired volumes of images elsewhere to be reviewed by “experts” is extremely important for high quality cost effective medical care. There are many clinics that are distant from specialty sites and would benefit from having access to expert consultations in this way.

Click on image to view full size

Figure 18. Surface rendering of 20 week conjoined twin fetuses. Note that they are facing each other, joined at the chest and abdomen, sharing a single heart (not shown).

In summary, 3D fetal imaging gives us far more control over the image and the subject we are imaging (in this case, the fetus) then we have had previously. (60,61) We can reorient this subject into standard planes, store large volumes of information to be reviewed and even “rescanned” later and perhaps even reevaluated from a completely different viewpoint than the acquisition plane at another time by a different sonologist. This technique is far more effective than videotaping since these volumes of information contain an infinite amount of views that can be reconstructed. This concept may have profound implications as well in the medical legal arena where volumes may be reevaluated by expert witnesses in completely different ways than they were originally viewed. Volumes contain an infinite amount of information compared to standard 2D displays. Ways in which 3D will impact our day-to-day life in fetal imaging have yet to be completely elucidated and we have only just begun to determine its effectiveness. My opinion is that 3D will not take the place of 2D (current standard fetal imaging) at least for the foreseeable future. 3D however will be important in many situations as an adjunct to 2D in the same ways that color Doppler or spectral Doppler is adjunct to gray scale. There is no doubt that 3D capabilities will be present and available on almost all ultrasound machines that are purchased in the future as more and more applications for the use of 3D will surface and it will become an important part of the ultrasonographers’ and sonologists’ armamentarium.