3D Ultrasound

5.4.6 3D/4D Ultrasound

3D US is a volumetric imaging technology that provides a 3D view of internal structures. Dynamic volumetric imaging, also known as “4D US” or “real-time 3D US,” extends the visualization with a time frame so that it is able to display motion instead of a static 3D data set.

Static 3D images can be acquired manually by moving a 2D transducer across a ROI, or automatically through the use of a 3D transducer that sweeps a 2D array of beams across the ROI. 3D/4D US requires rapid automatic sweeps of multiple adjacent 2D cross-sections.

Software

The software is the core of volumetric imaging technologies, especially 4D visualizations, which need highly optimized algorithms. For applications like scanning a heart, a gated technique is applied to avoid distortion of a 3D data set due to movement. Tomographic images are rearranged according to the phase of the cardiac cycle and a 3D data set is constructed with only tomographic images at the same phase of the cardiac cycle. The heart can be seen beating three-dimensionally by reconstructing many 3D data sets into a single cardiac cycle.

Strengths and Weaknesses

3D images provide examiners with an abundance of information, reducing the amount of interpretation needed and limiting the probability of misdiagnoses.

Compared to common US modifications, the amount of data involved is much higher. The depiction of one ROI demands up to 20 GB of data storage. Despite the increasing capability of computers, the processing time of data sets is still a limiting factor.

Recent Developments and Current Research

US travels through soft tissue at an average speed of 1540 m/s, which limits 3D scanning speed. The parallel receiving technique uses one broad US beam that is transmitted; its echoes are received as plural ultrasonic beams. In a 2D array probe, a high degree of parallel receiving is used and high-speed 3D scanning is possible. As a result, the profound advancements in 3D/4D imaging are mainly due to a general evolution of electronics and transducer arrays from linear systems to 1.25D, 1.5D, 1.75D, and 2D arrays and the latest matrix phase transducers, which are a current field of research [58].

Currently, there are plans to make 4D US available through handheld devices, which has already been achieved in high-end devices [59].

Today, the relevance of 3D/4D US systems is still low due to high purchasing costs and technological performance issues. In the future, the relevance of these systems is expected to rise as a result of technical improvements and mass-market adoption.

CartoSound

A 3D ultrasound image is acquired by a three-coil, electroanatomic mapping sensor and locator within the tip of the ICE catheter (SoundStar 3D catheter) and displayed with CartoSound software (Biosense Webster). The location and the direction of the phased-array image fan, with characteristic pitch, yaw, and roll components of catheter-tip motion, are displayed within the x-, y-, and z-axes. Three-second segments of two-dimensional (2D) ICE images are acquired during ECG gating. A single gated image is displayed on the ultrasound viewer, and the endocardial contour is manually or automatically drawn. By repeating this process, a series of contours in the chamber is acquired and is collated into a complete volume depicting point-to-point component interpolation and displayed.

The merit of the system is real-time acquisition of the anatomic information during the procedure. An ICE can be used for the transseptal puncture, location of the esophagus, location of PV–left atrium junction, view of the lasso (or spiral) catheter position in the veins, and monitoring for possible pericardial effusion development during the procedure. The validation study in animals using CartoSound showed that the actual anatomy and 3D ultrasound sites were only divergent by 2.1 ± 1.1 mm for atrial and 2.4 ± 1.2 mm for ventricular sites.¹¹ Using the CartoSound system, potential errors for CartoMerge approach can be corrected (Fig. 8-7).

Acquiring and Aligning a Volume

Successful 3D US first requires optimization of the two-dimensional (2D) image. The correct starting insonation angle with regard to the structure(s) of interest is of paramount importance. Not all sections that can be reformatted (extracted) from a 3D volume have the same resolution. In diagnostic US, the axial resolution (in the axis directed perpendicular away from the transducer) is better than the lateral resolution. Therefore the resolution in the acquisition plane (also called azimuth plane) is best; this plane is typically displayed in panel A directly after volume acquisition. The so-called range plane (blue plane in Fig. 173.2A), which cannot be obtained in real-time scanning by simply rotating the transducer, offers the lowest resolution, yet it often contains valuable information that cannot be obtained by maneuvering the transducer on the maternal abdomen, or in case of transvaginal scanning, in the vagina.

The starting plane for a volume acquisition should be placed to show the structures of interest as well as possible. Then the volume is acquired and initially displayed in the multiplanar mode (showing three orthogonal planes: A, B, and C; Fig. 173.6). Each of these planes can be aligned, using rotation around the X, Y, and Z axes, respectively, to achieve the intended orientation, typically in standard anatomic sectional planes. A clinical example of this workflow for the fetal face is shown in Fig. 173.2.

3D ultrasound

The introduction of 3D ultrasound into gynecological consultations has yielded improvement in the diagnosis of Müllerian malformation (Grimbizis et al., 2013) of the uterine cavity. With 3D echo we are able to visualize the coronal uterine plane, which aids the diagnosis of these malformations. The gold standard for diagnosis is magnetic resonance imaging (MRI), but 3D ultrasound presents high-quality images with a sensitivity, specificity, PPV, and NPV of 83%, 100%, 89%, 100%, respectively, comparable to MRI (Saravelos et al., 2016a).

3D ultrasound also enables calculations of endometrial volume, assessment of uterine vascularization, or assessment of the severity of adenomyosis (penetration and invasion by the glands and endometrial stroma within the myometrium). While attempts have been made to establish a relationship between endometrial volume and the pregnancy rate by embryo transfer, the results are contradictory. Some reports suggest that volumes less than 2 mL or greater than 8 mL are related to worse pregnancy outcomes, but other studies do not confirm these results. In fact, when the endometrial thickness is greater than 6.5 mm in the proliferative phase, the endometrial volume measured through virtual organ computer-aided analysis (VOCAL) does not show differences in the rate of evolutionary gestation in oocyte donation cycles, although those patients with very reduced endometrial volume seem to have worse reproductive results (Labarta et al., 2017). Studies of uterine vascularization (assessed by 3D ultrasound) in relation to pregnancy rate are inconclusive (Chen et al., 2017b). In the case of adenomyosis, there is increasing evidence of its negative effect during embryo implantation and perinatal outcomes (Puente et al., 2016).

INTRODUCTION

Three-dimensional (3D) ultrasound has become the most rapidly evolving technique in fetal imaging, but some examiners are still using 3D and four-dimensional (4D) techniques only to demonstrate the fetal face to the parents, which has made this new technique very popular with them as well. However, the concept of ‘volume ultrasound’ introduced a few years ago enabled a more comprehensive medical and clinical application of 3D technology to prenatal diagnosis.¹ A volume data set of the region of interest is acquired digitally, and the information stored can be displayed in different ways to highlight the spatial arrangement of a specific structure in the region of interest. Many colleagues may still be unfamiliar with all of these features, which are now well established in targeted prenatal diagnosis for ruling out or clearly demonstrating fetal malformations. In this chapter we will review the potential of volume ultrasound and the application of some display modes in clinical work.

CHAPTER 14 THREE-DIMENSIONAL AND FOUR-DIMENSIONAL ULTRASOUND APPLICATION IN PRENATAL DIAGNOSIS

3D imaging

The widespread acceptance of 3D ultrasound in obstetrics and gynaecology was helped considerably by the development of such transducers since they do not require any movement relative to the investigated tissue during acquisition. A single volume obtained at rest with an acquisition angle of 70 degrees or higher will include the entire levator hiatus with symphysis pubis, urethra, paravaginal tissues, the vagina, anorectum and pubovisceral muscle from the pelvic sidewall to the posterior aspect in the area of the arcus tendineus of the levator ani (ATLA) to the posterior aspect of the anorectal junction (see Fig. 5.20). The levator hiatus as seen on translabial 3D/4D ultrasound or MRI is the plane of minimal dimensions between the symphysis pubis/pubic rami anteriorly and the pubovisceral or puborectalis muscle laterally and posteriorly. Hiatal dimensions can be measured both in the axial plane (Dietz et al., 2005b) and in a rendered volume (Dietz et al., 2011c). The latter method is easier and at least as reproducible and may be preferable, given the non-Euclidean (warped) nature of the plane of the hiatus (Kruger et al., 2010).

The main advantage of volume ultrasound for pelvic floor imaging is access to the plane of the levator hiatus, i.e. the axial or transverse plane. Up until recently, pelvic floor ultrasound was limited to the midsagittal plane. Parasagittal (see Fig. 5.19) and coronal plane imaging (see Fig. 5.20, top right, for an example) may at times be helpful, although there are no obvious points of reference. Imaging planes on 3D ultrasound can be varied in a completely arbitrary fashion in order to enhance the visibility of a given anatomical structure, either at the time of acquisition or offline at a later time. The three orthogonal images (i.e., three planes at right-angles to each other – sagittal, transverse and axial) are complemented by a ‘rendered image’, i.e., a semitransparent representation of all volume pixels (voxels) in an arbitrarily definable ‘box’. The bottom right-hand image in Figure 5.20 shows a standard surface rendered image of the levator hiatus, with the rendering direction set from caudally to cranially, which is most convenient for imaging of the levator muscle. Midsagittal, axial and coronal views of the levator hiatus are given in the ‘orthogonal’ images in the top row and bottom left.

Three-Dimensional Imaging

Over the last decade, three-dimensional (3D) ultrasound has revolutionized the way clinicians diagnose, provide treatment, and manage patients with a variety of anomalies. 3D ultrasound is a valuable tool used in various fields, including applications in cardiac, obstetric, abdominal, and vascular studies. At present, it is possible for 3D ultrasound to construct high-resolution volume images that are comparable with those produced by computed tomography (CT) and magnetic resonance imaging (MRI) modalities at a lower cost and within a shorter time interval.

In some clinical scenarios, 3D imaging is the modality of choice used for direct visualization in defining precise anatomic locations and pathologies, as well as becoming indispensable in guiding surgical and percutaneous procedures. These advancements are, in part, because of the transducer design, ease of workflow, and computer software algorithms. The examinations are performed with a single transducer made up of a matrix array technology constructed of thousands of active elements providing performance in two-dimensional (2D), 3D, and four-dimensional image planes. It is important to recognize that a 2D transducer will only show one image plane at a time. However, an advantage to using a 3D transducer is its ability to simultaneously capture multiple image planes.

Moreover, the 3D transducer can be used in the biplane mode that displays two separate 2D planes. Biplane imaging allows for the operator to target the reference plane, view the region of interest, and move onto the second plane to observe an additional planar image (Fig. 4.16). Once this region has been identified, the operator can acquire a single image or cineloop.

The two most commonly used 3D techniques are multiplanar reformatting (MPR) and volume rendering (VR). When initiating the MPR dataset, the screen will be divided into four quadrants, three of which represent the X, Y, Z axes or orthogonal planes (i.e., longitudinal, transverse, and coronal) with the fourth representing the rendered image or cineloop. Once the dataset has been acquired, it can then be manipulated on the ultrasound system or off line on a picture-archiving and communication system with quantification software the same way CT and MRI images are viewed. During the postprocessing phase, the acquired images captured in the X, Y, Z, and slice MPR planes, as with cineloops, can be cropped in different axis planes to define the anatomic structures and pathology in question (Fig. 4.17). In addition, images acquired with color flow or power Doppler can be cropped, rotated, and further optimized with various postprocessing features such as gain, steering options, zoom, and chroma tint; use of these features will allow for the acquisition of precise images.

Although 2D duplex ultrasound is a reliable imaging modality, there are clinical scenarios when more information is required. For example, when obtaining 2D measurements for organs, the data gathered can vary and be inaccurate at times because of anatomic boundaries. With 3D imaging, the operator has the ability to acquire datasets via simplified volume acquisition, hence having more options and flexibility to manipulate the transducer, view the desired anatomic image planes, and define the potential segments. As a result, the operator captures a complete examination with better qualitative and quantitative information, reducing the overall study time and allowing for more accuracy in making the clinical diagnosis.

In the clinical setting, 3D ultrasound can be used during a variety of clinical scenarios as follows: when evaluating the severity of vessel obstruction, filter patency and position, wall stent abnormalities, stenotic lesions and aneurysmal segments. Furthermore, 3D ultrasound aids in confirming potential tumor invasion of adjacent vascular structures, defining areas of tortuosity and kinking, determining characteristics of vascular lesions, evaluating for suspected venous compression, and defining vascular organ perfusion.

In conclusion, it is evident that 3D imaging modality offers a multitude of advantages. It plays a role in assessing for different types of deep and superficial venous pathology, and because of its affordability compared with cross-sectional imaging, could allow for mainstream use in clinical practice setting. A particular benefit for the endovascular specialist is its usefulness in planning a course of action before an invasive procedure. As this technology continues to be developed, we should expect that 3D ultrasound imaging would eventually become part of the mainstay in the clinical setting.

Artefacts in three-dimensional scanning

Some artefacts are only encountered in three-dimensional (3D) ultrasound.³⁸ They may stem from errors in the data collection process, for example movements of the probe that are too rapid result in gaps in the final image that may be depicted as such in registered 3D but when the data are collected freehand (where no correction is made for the z-plane movement), they produce errors in the geometry of this axis that are not necessarily apparent in the image: this is why measurements from this type of unregistered 3D scan are unreliable.³⁹ Obviously errors in the system that generated the spatial coordinates of the raw images distort the final rendered image.

Speckle in the ultrasound data also produces defects in the final image and they are particularly obvious in surface-rendered displays where they show as serpiginous holes.⁴⁰ Though unaesthetic, they do not affect the validity of the image; speckle reduction techniques can be used to mitigate the problem.

If the rendering algorithm does not interpolate adequately for missing slices, a ‘gappy’ rendered image results giving a toothcomb appearance when the three-dimensional set is rotated to align the z-axis.

If the structure being imaged moves during the data collection, the final rendered set will be distorted. This is particularly a problem with an active fetus and produces bizarrely distorted three-dimensional images. It is also a problem with vascular three-dimensional imaging because of vessel pulsations.

Display Modes

Figure 9-22 demonstrates the two basic display modes used on 3D ultrasound systems. The multiplanar or orthogonal display mode shows cross-sectional planes through the volume in question. For pelvic floor imaging, this most conveniently means the midsagittal, the coronal, and the axial or transverse plane.

One of the main advantages of volume ultrasound for pelvic floor imaging is that the method gives access to the axial plane. Until recently, pelvic floor ultrasound was limited to the midsagittal plane.^9,101,102 Parasagittal and coronal plane imaging have not been reported, perhaps because there are no obvious points of reference, unlike the convenient reference point of the symphysis pubis on midsagittal views. The axial plane was accessible only on MRI; Figure 9-23 provides an axial view of the levator hiatus on MRI and 3D ultrasound.¹⁰³ Pelvic floor MRI is an established investigational method, at least for research applications, with a multitude of papers published over the past 10 years.^104–113

Imaging planes on 3D ultrasound can be varied in a completely arbitrary fashion to enhance the visibility of a given anatomical structure at the time of acquisition or offline at a later time. The levator ani, for example, usually requires an axial plane that is slightly tilted in a cranioventral to dorsocaudal direction. The three orthogonal images are complemented by a rendered image, which is a semitransparent representation of all voxels in an arbitrarily definable box, termed the Region of Interest (ROI). Figure 9-22 (bottom right image) shows a standard rendered volume of the levator hiatus, with the rendering direction set caudally to cranially, which seems to be most convenient for pelvic floor imaging. The possibilities for postprocessing are restricted only by the software used for this purpose; programs such as GE Kretz 4D View (GE Medical Systems Kretztechnik, Zipf, Austria) allow extensive manipulation of image characteristics and output of stills, cine loops, and rotational volumes in bitmap and AVI formats.