Three-dimensional (3D) volume imaging of the female pelvis is one of the most important advances in women’s imaging within the last decade.^1,2 Volume imaging is not new to radiology, as it has been used extensively as part of computed tomography (CT) and magnetic resonance (MR) imaging for several decades. Sonographic imaging, however, has been relegated for more than 30 years to individual acquisitions of thin two-dimensional (2D) tomographic image slices that are operator dependent and time consuming to obtain.³ The quality of the individual 2D images are largely dependent on the expertise of the operator that produces them, and even a careful review of these individual “snapshots” will often fail to demonstrate pathology if the operator did not perceive the abnormal finding.

More recently, 3D sonography has enabled us to move sonographic imaging into a new era of rapid, automated, and comprehensive imaging and displays.^1-4 Images of the entire pelvis can be accomplished with only three volume acquisitions: one of the uterus and one of each ovary.⁴ These volume acquisitions typically contain all of the sonographic anatomic information within the pelvis, thus affording improved perception of the overall global anatomy. They can subsequently be rescanned electronically in any plane by a different operator who may be off site.^1-4

The earliest 3D ultrasound probes required manual sweeps through a volume of interest, typically in the sagittal plane (Figures 49-1 and 49-2). This “freehand” method collects a series of consecutive 2D slices. The current methods of sonographic volume acquisition utilize automated transducers, which have a dedicated 3D probe within their housing, mechanized to gather 2D sonographic information from one side of the probe to the other. The operator holds the transducer still while the probe sweeps from one side of the housing to the other, thus acquiring the volume in a smoother, more systematic and reproducible fashion in less than 15 seconds. However, the faster the speed of this mechanical acquisition, the lower the resolution of the images. Different types of processing can also modify the quality and appearance of the volume data, such as the use of harmonics or other types of image-modification settings.

Once the 3D volume has been acquired, this large amount of sonographic information can then be presented in many different ways. Software programs can reconstruct raw information into axial, sagittal, and coronal views, no matter the original scanning plane.^1,2,4-6 A coronal view of the female pelvis is the most accurate and informative view for diagnosing uterine and tubal pathology, though the reconstructed coronal images will never achieve the resolution of the original plane. The 3D images are processed and displayed on a monitor in a multiplanar format showing the three (X, Y, and Z) orthogonal planes with a single dot at their intersection. The operator can navigate easily through the volume using the four basic tools of X, Y, and Z rotation and slicing from one end of the volume to the other.⁷ Volumetric quantitative measurements of portions of the anatomy can then be made within the data set.⁸

Volumes of 3D information can also be displayed using surface rendering techniques, where fluid is used as an interface to delineate the surface of an organ.⁹ If there is fluid within the volume, a rendering line of sight can be drawn through that volume, close to the interface with the solid area, such that the topographical surface of the solid area can then be displayed. The surface-rendered image is a thicker slice through the volume with depth perception improved by different computer-generated shading and lighting effects. Alternatively, multiple tomographic cuts that are equidistant from each other can also be generated to display the information within the entire volume, much like a CT or MR display.¹⁰ To accomplish the multiple tomographic cuts, a single plane is selected from within the volume, and the entire volume is sliced in that orientation. The volume can be displayed in an inverse mode, in which the solid areas “melt away” and all of the cystic areas are made into a cast of the inside of the volume.^11,12 Finally, the acquisition of the volume can also be made using color Doppler, such that 3D blood flow data can be incorporated into the information.¹³ One can then interpret the Doppler image qualitatively, in terms of mapping, or quantitatively, in terms of blood flow velocity and waveform.^13-16

All of these types of displays, and more, are used to tease out the valuable raw sonographic information within a volume, regardless of how the volume was acquired. This makes 3D volume imaging of the pelvis far more comprehensive compared to the traditional acquisition of individual 2D sonographic images.^1,2 The volume is also obtained faster, thus promoting greater patient comfort and safety. Once the volume has been acquired, it can be sent to a facility distant from the scanning site to be reviewed by several different practitioners, as though the patient had been sent to that location.^1,2,4 The raw data captured within these volumes permits sonography to emerge from an era of operator dependency by permitting the acquisition of an entire area of interest at the time of the scan. This increased automation allows operators of varied experience levels to collect the same data without depending on their subjective recognition of abnormalities. Let us now further explore the applications and technical considerations of some of the aforementioned displays.

Surface Rendering

When there is sufficient fluid within a volume acquisition, an interface is formed between the fluid and solid surfaces, permitting a surface-rendering display (Figures 49-3, 49-4, 49-5, 49-6). For example, in patients having sonohysterography, one can display the surface of polyps or fibroids abutting the saline-filled cavity, much like a virtual hysteroscopy. This technique allows us to “enter” fluid-filled cavities within the volumes, much like using a scope to directly visualize the surface of these lesions.⁹ We can use the same technique when visualizing the surface of the bladder in patients with bladder tumors or other mucosal abnormalities.^17,18 The internal aspect of an ovarian cyst can also be evaluated with surface rendering to show areas of nodularity or polypoid lesions in cases of complex cystic masses such as ovarian neoplasms.

When using the surface-rendering technique, it is often helpful to utilize software with an electronic scalpel to carve away portions of the images that are superfluous to the anatomy of interest. The ability to cut away these unwanted echoes from in front of the target area is best accomplished by rotating the rendered image in different directions to identify the best cutting plane. One can even take an entire volume that contains a cyst and slice it in half with the electronic scalpel to pry the two parts apart and see the rendered inner surface of the cyst inside.

Finally, using the surface-rendering mode, the practitioner can render just a thin portion of the volume, thus discarding the rest. This thick-slice technique results in a reconstructed image that comprises a thicker slice of anatomy than standard 2D imaging. This technique emphasizes the edge enhancement of structures and provides a contrast-rich image.

Inversion Displays

3D inversion display is a method by which the volume is rendered by forming a cast of all of the cystic areas within the volume and having the solid areas become lucent such that only the cast is visible (Figures 49-7, 49-8, 49-9).^11,12 This provides a view of the entirety of the cystic areas within the volume all at one time. The volume can be rotated in any way necessary to view all of the aspects of these cystic portions. An example of how this can be useful is in the evaluation of follicles within an ovary. All of the follicles can be made opaque within an ovary and counted and measured simultaneously. One can image an ovary that has polycystic ovarian syndrome, revealing the characteristic appearance of the cluster of small follicles along the periphery of the ovary. The evaluation of hydrosalpinges lends itself particularly well to the inverse mode, since the fluid-containing tube may enter multiple different 2D planes. The inverse mode is sometimes the only way to display the entire structure because there is often not a single plane that will demonstrate all permutations of the ectatic tube. One can also do an inverse-mode image of the endometrial cavity containing fluid, such as during a sonohysterogram. This technique gives the appearance of positive contrast within the uterus and the image produced is similar to a hysterosalpingogram.

Tomographic Ultrasound Imaging

Tomographic ultrasound imaging (TUI) is another method by which the information within a volume can be displayed (Figure 49-10).⁴ Multiple reconstructed 2D cross-sections of the volume are displayed as static images or in a cine loop. This brings sonography closer to other forms of cross-sectional imaging, such as CT and MR. The pelvis can be imaged by acquiring three or four volumes (at least one of the uterus and one of each ovary), and then displaying them using TUI so that they can be read as still images offline or at a site distant from the patient. One can also navigate through the volume and do a virtual rescan of the patient if necessary.⁴ Benacerraf et al studied the feasibility of performing pelvic sonograms using this technique and found that pelvic imaging could be done very accurately and more efficiently using 3D sonography displayed in this way.⁴ The accuracy of the scan read offline was similar to the original reading of the standard 2D sonogram, done at the time of the patient’s visit. Thus, TUI could help us make great progress toward standardizing and optimizing the efficiency of pelvic sonography in the future.

Doppler Sonography in 3D

Doppler evaluation of blood flow using 3D displays can provide two important pieces of information (Figures 49-11, 49-12, 49-13, 49-14, 49-15). First, the evaluation of the morphology of the vessel tree provided by 3D power Doppler volumes aids in identifying malignant ovarian masses.¹⁹ In particular, it can display the characteristic appearances of tumor vessels, such as the areas of focal stenosis, irregularity, and aneurysmal dilatation of blind-ending vessels seen in neoplasms.¹⁹ There are several studies showing that 3D sonography with color Doppler can help to depict the overall vessel density and branching patterns within tumors.^13,19,20 The actual color mapping of the vascular architecture within the lesion is of particular importance and can display the characteristic cluster of irregular tumoral vessels in the center of a malignant ovarian tumor. The presence of central intratumoral vascularity had a 90% positive predictive value in a study by Fleischer et al for predicting malignancy. Conversely, the absence of such central vascularity had a high negative predictive value of 96%.¹³ In another study, 3D power Doppler imaging was more accurate than 2D sonography at defining the morphologic and vascular characteristics of ovarian tumors.²⁰ Although all malignancies in this group of 71 total women were correctly identified with both 2D and 3D standard imaging, the specificity increased by adding 3D power Doppler, thus improving the diagnostic accuracy. In another study of 77 benign tumors, 6 borderline tumors, and 21 invasive malignancies, the vascular features differed significantly between benign and malignant tumors on 3D sonography.¹⁹

3D color Doppler is also used to assess the vascularity of the uterus and the endometrium.^21,22 Fleischer et al reported that 3D color Doppler provides valuable information before and after fibroid embolization.²¹ Among 20 patients who had a total of 31 fibroids studied, the maximum reduction in vascularity occurred 1 day after the procedure. The volume reduction was maintained at the 3-month follow-up examination. The hypervascular fibroids tended to shrink more than those that were isovascular or hypovascular. The depiction of fibroid vascularity by color Doppler also improved the ability to determine the size, location, and extent of myometrial involvement. Combined, this information helps to stratify patients and predict treatment success.

The second major contribution of Doppler imaging within 3D volumes is the ability to quantify the number of vessels and volume of flow within an area of interest. Jokubkiene et al studied the changes in volume and vascularization of normal ovaries during typical menstrual cycles using 3D power Doppler.²³ They showed that vascular indices in the dominant ovary and the dominant follicle/corpus luteum increased during the early part of the cycle, with the vascular flow index 1.7 times higher on the day before ovulation than 4 days before ovulation. The vascular flow index in the corpus luteum was 3.1 times higher 7 days after ovulation than in the follicle on the day before ovulation.²³ This study used volumetric sonography to successfully demonstrate that substantial changes in the vascularization of the dominant ovary take place during the course of the menstrual cycle, a phenomenon potentially helpful when evaluating patients with infertility as well as those with ovarian masses.

Investigators have also used 3D power Doppler as a tool to study the changes in endometrial and subendometrial volume and vascularity during the normal menstrual cycle.²² The endometrial and subendometrial vascularity appears to increase throughout the follicular phase, decreasing after follicular rupture, only to increase again during the luteal phase. 3D endometrial volume, described later in the chapter, and power Doppler can predict endometrial carcinoma and hyperplasia, particularly in patients with post- and perimenopausal bleeding.

There are several novel and highly reproducible methods for quantifying the vascularity within a sonographic volume.^14,16,24-28 This can be done using a 3D histogram, where the distribution of color percentages and flow amplitudes can be measured within the volume. When a 3D volume is obtained with color-flow information, the area of interest can be delineated so that the vascular parameters can be calculated within a specified box.¹⁵ The vascularization index (VI) measures the percentage of color voxels within the box and basically provides a vessel count. The flow index (FI) is a measure of the flow traversing the region at the moment the volume was acquired. The vascular flow index (VFI) combines the information from the previous two measurements and provides a measurement of the blood flow and vessel count within the selected volume. Software programs can facilitate calculation of these values by generating histograms in delineated areas of interest.

Alcázar et al used Doppler vascular sampling with these quantitative methods to predict the presence or absence of ovarian cancer within vascularized complex adnexal masses.¹⁴ They analyzed the vascularization of highly suggestive areas with gross papillary projections or solid areas using the vascular indices (vascularization index, flow index, and vascular flow index), which were automatically calculated for these areas within the volume. The median vascularization index (MVI), FI, and VFI were significantly higher among malignant tumors compared with nonmalignant lesions.¹⁴ While these investigators demonstrated no differences in the resistive index, pulsatility index, or peak systolic velocity in benign versus malignant tumors, several other studies showed significantly higher values for the latter.^{14,16,24,25,27} 3D color Doppler used to evaluate for cervical carcinoma indicates a significantly higher color score and lower resistive index for tumors 4 cm or larger.^29-31 Conversely, another study by Alcázar et al suggested that 3D power Doppler sonography did not significantly add specificity or sensitivity when compared to 2D power Doppler; therefore, the concept remains somewhat controversial for use in clinical practice.²⁶ Further categorization of tumors in the future may better delineate other consistent, characteristic index patterns.

Conventional 2D sonography is known to have many potential artifacts that are easily recognizable in clinical practice (Figure 49-16). Solid or calcified masses can cast an acoustic shadow or attenuation of the beam. Cystic areas produce through-transmission that is easily recognized. There are also important color and power Doppler artifacts to consider, such as aliasing and motion.

Although these artifacts are fairly standard and easily recognized using 2D standard sonography, they are often confusing and can even be compounded within a volume. A shadow or an area of sound enhancement is recognized as such because the source of the artifact is usually seen in the same 2D image. Once an image is reconstructed from a volume, the artifact will be visible, but often without the origin of that artifact, making it much harder to recognize. For example, a shadow may be difficult to interpret on an image where the calcification or device (eg, IUD) that caused that shadow may only be seen in a different plane. The shadow may be mistaken for pathology if the source of the artifact is not seen. Doppler artifacts such as those relating to gain, aliasing, and flash may be seen within reconstructed images on a plane different from the acquisition view. It is therefore important to go back and navigate through the original volume in the orientation of acquisition to determine the source of shadows and unusual findings on images created from volumetric acquisitions. There are also artifacts that are unique to 3D volume acquisition and reconstruction, which include motion (of heart, limbs, bowel, etc) artifacts and shadows from objects located adjacent to the area of interest but not visible in the volume obtained.

3D volume acquisitions provide the opportunity to make accurate volume measurements of contained regions. One in vitro study demonstrated that volume measurements of balloons of different shapes (as small as 20 mL and as deep as 10 cm in a tissue-simulating fluid bath) were highly accurate when using 3D sonography, significantly surpassing the accuracy of 2D sonography.³² This and other investigations have also demonstrated the intra- and interobserver reliability of in vitro calculations done within volumetric sonography data sets.^32,33 In one study, volume measurements performed on a 3D sonography data set were highly reliable to within 4% of true volumes.³³ Studies in vivo indicate that 3D volume measurements are also accurate when comparing, for example, the measured volume within a urinary bladder with the voided urine volume,³⁴ a potentially important measurement for urodynamic studies. 3D sonography is particularly valuable for measurement of irregularly shaped organs that must traced in multiple planes.^32,34 These measurements can be made offline, within a stored volume.³²

Virtual organ computer-aided analysis (VOCAL) is a software program that allows the practitioner to trace the contour of a region of interest within an acquired volume, stepwise over the course of 180 degrees, thus enabling the calculation of the volume of the selection and other indices within the volume (eg, vascular flow index).^35,36 This permits the practitioner to highlight even irregularly-shaped portions of the organ of interest so that the volume is measured accurately. VOCAL provides a reproducible method of estimating, for example, the endometrial volume.³⁵ The volume of the endometrium may be a more accurate method of assessing its health than the standard single-width measurement made on either 2D or 3D sonogarphy.³⁵

Volume Measurement and Endometrial Cancer

Gruboeck et al postulate that total endometrial volume serves as a better predictor of endometrial hyperplasia than focal endometrial thickness. The authors found that endometrial thickness had a sensitivity of 83.3% for diagnosis of endometrial cancer, while endometrial volume measured using 3D sonography had a sensitivity of 100% (mean volume of 39 mL vs 0.9 mL).³⁷ Measurement of irregular volumes such as the endometrium would be very difficult without 3D sonography and software programs such as VOCAL.

Staging of endometrial cancer is, in part, guided by depth of invasion into the myometrium. The thickness of cancer-free myometrium measured on 3D sonography correlates well with actual measurements in histologic specimens, surpassing estimates from 2D studies.^38,39 Although limited by small sample size, a recent prospective study could not claim superiority of 3D sonography over MRI at detecting invasion preoperatively, though sonography possessed greater specificity.⁴⁰ Endometrial blood supply travels from the serosal edge of the uterus toward the cavity, thus increased low-resistance flow in radial arteries can be seen on coronal 3D color Doppler sonograms in the myometrium adjacent to the cancer.^41,42

Four-dimensional (4D) sonography realistically represents 3D objects as they move through space in real time.¹ Most current instruments still employ mechanical 3D transducers, which tend to produce choppy 4D recordings as the inside of the probe sweeps from one end of the transducer housing to the other, sequentially acquiring many 2D images. In such cases, the resolution of the images is inversely related to the rapidity of the frame rate, thus making more “fluid” collections of images appear less crisp. It is important for future instruments, therefore, to eliminate the mechanical aspect of volume imaging and replace it with the ability to obtain instant volume acquisitions with improved image resolution and smoother real-time capabilities.¹ This is currently accomplished with recently available matrix array probes that consist of thousands of individual subelements that each send out their own ultrasound beams. The individual images from these subelements are almost simultaneously combined and reconstructed to produce a volumetric image in real time without any movement of the internal hardware. 4D sonography can, furthermore, be combined with color Doppler. Current applications of this technology mostly relate to fetal cardiac imaging, but also to studies of the liver and tumors.

The use of 3D volume displays in conjunction with sonohysterography can potentially enhance the accuracy of evaluation of the uterine cavity (Figures 49-17, 49-18, 49-19).^43-45 Anechoic saline introduced into the endometrial cavity before volume acquisition juxtaposes the distended uterine walls, outlining and separating more echogenic polyps and submucosal fibroids. In practice, however, controversy remains regarding the extent to which 3D sonohysterography adds significant additional information when compared to 2D sonohysterography. Alcazar et al provide an excellent summary of studies comparing 2D and 3D sonohysterography implemented in women suspected of having various intracavitary lesions. In general, studies were limited by smaller sample sizes and lack of 3D data from patients who had normal 2D studies.⁴⁶