Clinical Utility of Three-dimensional US
Abstract
Three-dimensional (3D) ultrasonography (US) is rapidly gaining popularity as it moves out of the research environment and into the clinical setting. This modality offers several distinct advantages over conventional US, including 3D image reconstruction with a single pass of the US beam, virtually unlimited viewing perspectives; accurate assessment of long-term effects of treatment; and more accurate, repeatable evaluation of anatomic structures and disease entities. In obstetric imaging, 3D US provides a novel perspective on the fetal anatomy, makes anomalies easier to recognize, facilitates maternal-fetal bonding, and helps families better understand fetal abnormalities. Three-dimensional pelvic US allows volume data sets to be acquired with both transvaginal and transabdominal probes. Viewing multiple 3D power Doppler US images in a fast cine loop has proved useful in angiographic applications. Three-dimensional prostate US can help make accurate volume assessments for dosimetry planning or for estimating prostate-specific antigen levels. In breast imaging, 3D US has the capacity to demonstrate lesion margins and topography, thereby helping differentiate benign from malignant masses. Three-dimensional US can also help determine the need for biopsy and help facilitate needle localization and guidance during biopsy. With recent advances in computer technology and display techniques, 3D US will likely play an increasingly important role in medicine.
Introduction
Conventional radiography merges three-dimensional (3D) data into two-dimensional (2D) summation images. Radiologists mentally reverse this process by forming 3D impressions of the underlying anatomy and disease. Tomographic data from ultrasonography (US), computed tomography (CT), and magnetic resonance (MR) imaging have made this mental reformatting process easier and more accurate. Recent advances in computer technology and display techniques promise to make medical imaging data even more accessible and clinically useful. Real-time 3D display, reformatting, and manipulation of US, CT, and MR images are now possible on inexpensive desktop computers.
Three-dimensional US is in the early stages of clinical assessment and adoption as it gradually moves out of the research laboratory and becomes commercially available for routine clinical use (,1–,6). In this article, we summarize pertinent data from earlier clinical studies of 3D US, describe the basic principles of this modality, and discuss the techniques used to produce 3D B-mode and Doppler US images. In addition, we suggest ways to optimize image quality and minimize artifacts. We also discuss potential clinical applications of 3D US.
Advantages of 3D US over 2D US
Two-dimensional US is a flexible, cost-effective imaging tool that allows users to see and record a large variety of thin anatomic sections in real time. However, conventional US has several disadvantages that 3D US has the potential to rectify. One of its major disadvantages is operator dependency. The operator sweeps the US beam back and forth across an organ many times while mentally integrating multiple 2D images into a 3D impression of the underlying anatomy and disease. This process is universally acknowledged as time-consuming and inefficient, and there is considerable interobserver variability. In contrast, 3D images can be reconstructed from data obtained with a single sweep of the US beam across the involved organ. Both the US information and the relative position of each tomographic section are accurately recorded. As a result, the exact relationship between anatomic structures is accurately recorded in the 3D image.
Another disadvantage of conventional US is the limited viewing perspective it allows. Sometimes, the patient's anatomy or position makes it impossible to orient the US transducer for optimal visualization of a clinically relevant relationship between two structures. Three-dimensional US allows unrestricted access to an infinite number of viewing planes.
In addition, 2D US is ill suited for monitoring the effects of therapy over a long period of time. To minimize artifacts, 2D US images are usually acquired with nonstandard patient positioning during various phases of respiration. To accurately assess the long-term effects of treatment during follow-up, the ideal would be to replicate the US images that best demonstrated the abnormality. Although it is usually possible to approximate an earlier image, one can never be sure if the changes on a subsequent image are substantive or merely reflect slight differences in imaging technique. Three-dimensional US allows comparison of two full data sets over time, thereby improving accuracy of evaluation.
Furthermore, with 2D US, a “flat” anatomic section is displayed on a video monitor or on film. With 3D US, different viewing algorithms allow the data to be displayed with a variety of techniques, including surface rendering, volume rendering, and multiplanar reformatting.
Finally, quantitative volume estimates made at 2D US are often based on images that are approximately orthogonal to each other, which may lead to inaccurate and variable results. Three-dimensional US has been shown to provide a more accurate and repeatable method of evaluating anatomic structures and disease entities (,7,,8).
Limitations of 3D US
All of the current 3D US data acquisition techniques are more cumbersome than conventional US techniques. Most involve adding a localizing technology or a mechanical motor to the transducer. The resulting assemblies are typically bigger and therefore more difficult for the user to manipulate. In addition, the data acquisition software usually requires considerably more user input than is required with conventional US. The larger data sets produced with 3D US also make data archiving and communication more challenging.
Different computer algorithms reconstruct data into 3D volumes at different speeds. Some produce images almost instantaneously, whereas others require several seconds to produce an on-screen image. Waiting for the 3D image to appear can be frustrating to users who are accustomed to having US images appear immediately.
The ability to view data with a variety of algorithms and from different perspectives may slow the image interpretation process. Inexperienced users may have to spend extra time finding the best algorithm and perspective for viewing the data. Many of the viewing programs require a considerable amount of image manipulation to obtain high-quality results.
Operators must be extremely careful when manipulating data with surface-rendering and volume-rendering algorithms because inappropriate settings may introduce artifacts into the image, thereby altering the diagnosis. We strongly advise reviewing the original raw data to clarify any suspicious findings on rendered images.
At conventional US, the operator can use a variety of techniques to increase the visibility of an organ. These include compressing the area being imaged and altering the patient's position and breathing pattern. These techniques may be more difficult to implement at 3D US. We use 2D US and have grown so accustomed to seeing angle-dependent artifacts behind structures (eg, shadowing behind stones or bowel loops) that we use the artifacts to help us make an accurate diagnosis. Three-dimensional US may allow a different perspective on a structure behind which an experienced diagnostician would expect an artifact. If the original US beam was introduced from a different direction, the expected artifact will not be present, and the operator may have less confidence in the diagnosis.
Techniques of 3D US
This section describes the various techniques of 3D US data acquisition, image reconstruction, and image display.
Data Acquisition Techniques
Three-dimensional images can be constructed with 2D US arrays, which produce 3D image data directly. More commonly, they are reconstructed from a series of 2D images produced with one-dimensional US arrays. Regardless of which method is used, one must know the relative position and angulation of each 2D image and must acquire the images rapidly or with gating to avoid motion artifacts. If these two criteria are not met, the 3D images may be inaccurate.
The four main types of 3D US data acquisition systems are (a) tracked freehand systems, (b) untracked freehand systems, (c) mechanical assemblies, and (d) 2D arrays.
Tracked Freehand Systems.—With tracked freehand systems, the operator holds an assembly composed of the transducer with an attachment and manipulates it over the anatomic area being evaluated (,,,,,Fig 1). Two-dimensional images are digitized as the transducer is moved. During this procedure, the exact relative position and angulation of the US transducer must be known for each digitized image, and the operator must ensure that there are no significant imaging gaps.
The three principal types of tracked freehand systems involve acoustic tracking (,,,,,Fig 1a), articulated-arm tracking (,,,,,Fig 1b), and magnetic field tracking (,,,,,Fig 1c, ,,,,,1d).
To date, the most successful tracked freehand technique is magnetic field tracking. This is a magnetic field system consisting of a transmitter that produces a spatially varying magnetic field and a receiver that contains three orthogonal coils that measure field strength (,,,,,Fig 1c, ,,,,,1d). The position and angulation of the receiver relative to the transmitter are determined by measuring the local magnetic field. To achieve accurate 3D reconstruction, electromagnetic interference must be minimized, the transmitter must be close to the receiver, and there should be no ferrous or highly conductive metals in the vicinity.
Untracked Freehand Systems.—With untracked freehand systems, 2D images are digitized as the operator moves the transducer with a smooth, steady motion (,,,,Fig 2). Although this technique is usually the most convenient for the operator, image quality is variable and depends largely on how smoothly and steadily the operator moves the transducer. To reconstruct a 3D image, a linear or angular space between digitized images is assumed. Geometric measurements such as distance or volume may be inaccurate and should not be taken because there is no direct information regarding the relative position of the digitized images.
Mechanical Assemblies.—With mechanical assemblies, the transducer is propelled or rotated mechanically, and 2D images are digitized at predetermined spatial or angular intervals (,,,,Fig 3). Although these mechanical transducer attachments are more cumbersome for the user, they also improve the geometric accuracy of the resulting images. To date, their greatest utility has been in intracavitary and intraluminal examinations, in which the area of interest is relatively small and motion artifact is less of a problem. Three different scanning techniques can be used: linear scanning, tilt scanning, and rotational scanning.
With linear scanning, a mechanical assembly moves the transducer in a linear fashion along the patient's skin (,,,,Fig 3a). This technique has certain advantages; for example, by tilting the transducer away from the vertical plane, 3D color and power Doppler US (which are angle dependent) can be performed. Linear scanning also allows adjustment of the interval between the 2D digitized images for proper sampling and makes 3D reconstruction very fast and efficient because the images are parallel and are separated by a predetermined interval.
With tilt scanning, the transducer is tilted about its face, and images are digitized at a predetermined angular interval (,,,,Fig 3b). The mechanism for tilt scanning is usually quite small, which allows easy handheld manipulations. However, because the digitized images are arranged like a fan, the space between them increases and the resolution decreases with increasing depth.
With rotational scanning, the transducer is rotated about its central axis (,,,,Fig 3c), producing digitized images in a propellerlike arrangement. As a result, the sampling distance increases and the resolution decreases as distance from the rotational axis increases. In addition, the digitized images intersect along the rotational axis, so that any motion creates artifacts at the center of the 3D image.
Two-dimensional Arrays.—With the first three types of data acquisition systems, mechanical motion is used to obtain 3D images. An alternative is to keep the transducer stationary and use electronic scanning with a 2D transducer array (,9), which can be square or circular. Such an array generates pyramidal or conical US pulses and processes the echoes to generate 3D information in real time (,,,Fig 4).
Image Reconstruction Techniques
The 3D reconstruction process involves the generation of a 3D image from a digitized set of 2D images. Two methods have been used for 3D reconstruction: a 3D surface model and a voxel-based volume model.
Three-dimensional Surface Model.—With a 3D surface model, the operator can outline the boundaries of the areas of interest on the 2D images manually or with a computer algorithm specifically designed to detect and outline these areas of interest. These boundaries can then be assigned a value or color to distinguish them from adjacent tissues, and a 3D surface model of the anatomy is built up and displayed. The main advantage of this approach is that it reduces the amount of 3D data needed; only information on a few boundaries or structures is required. This leads to shorter 3D reconstruction times and greater efficiency. Another advantage is that contrast between structures is artificially increased. However, this increase in contrast may also distort or misrepresent subtle image features and may cause loss of valuable information, particularly in areas that have subtle tissue differences. In addition, manual identification of boundaries can be tedious and time-consuming.
Voxel-based Volume Model.—With the more popular voxel-based volume model, the computer builds a 3D voxel-based volume (3D grid) as it places each digitized 2D image into its correct location in the volume. This process preserves the original information during 3D reconstruction and allows a variety of rendering techniques. Unfortunately, this method also generates very large data files, which slows processing and requires large amounts of computer memory.
Image Display Techniques
The ability to detect disease on 3D images also depends on using the appropriate rendering technique. The four basic types of rendering techniques currently in use are surface rendering, multiplanar reformatting, combined surface rendering and multiplanar reformatting, and volume rendering.
Surface Rendering.—With surface rendering, the operator identifies the boundaries of pertinent structures either manually or with an algorithm, after which the computer creates a wire-frame representation of these regions. The boundaries are then shaded and illuminated, which gives shape to the surfaces of the structures.
Multiplanar Reformatting.—Once a 3D voxel-based volume has been reconstructed, the imaging information can be viewed with two multiplanar reformatting techniques. One popular technique presents the operator with three orthogonal planes taken from the volume. Computer-user interface tools allow the operator to rotate and reposition these planes so that the entire volume of data can be examined. The principal advantages of this technique are its speed and simplicity.
With the second technique (texture mapping), the 3D image appears as a polyhedron. The appropriate imaging plane is “painted” on each face of the polyhedron, which can be rotated to obtain the desired image orientation, and any of the faces can then be moved in or out (ie, sectioned). The advantage of this technique is that the operator can always relate the manipulated plane to the rest of the anatomy (,Fig 5).
Combined Surface Rendering and Multiplanar Reformatting.—It has recently become possible with modern technology to view part of the 3D US data with multiplanar reformatting and the remainder with a surface-rendering technique.
Volume Rendering.—Volume rendering casts “rays” through the 3D voxel-based volume image and projects the results onto a 2D plane (,Fig 6). A wide spectrum of visual effects can be produced depending on how the algorithm interacts with each voxel encountered by a particular ray. For example, if each voxel value is multiplied by 1 and the values are summed, a radiograph-like image can be produced. The voxel values can be multiplied by selected factors and summed to produce varying degrees of translucency. Maximum-intensity-projection images are produced when only the voxel with the maximum intensity along each ray is displayed. Optimal results are obtained when there is good contrast between tissue and adjacent structures (eg, fluid).
Further Imaging Considerations
Most 3D US images represent a compromise. From a diagnostic perspective, it is generally desirable to obtain as large a data set as possible because showing the maximum amount of normal tissue surrounding a lesion generally aids in diagnosis. For example, the more of the liver that can be included in a scan, the easier it will be to assign a lesion to a specific Couinaud segment. In addition, higher-resolution 3D images allow more detailed depiction of structures within a lesion or organ. If an area is undersampled, subtle anomalies may be missed.
However, with larger data sets come greater technical challenges. Larger data sets typically require more computer memory, more short- and long-term computer storage capacity, and faster processors to display, manipulate, and interpret the data. Interfaces must be devised that allow large volumes of data to be viewed and interpreted in a fast, intuitive manner. In addition, larger data sets are typically more prone to motion and reconstruction artifacts, and transmitting them over networks is slower and more difficult.
Imaging Artifacts
By understanding the origin and appearance of imaging artifacts, examiners can minimize the number of artifacts and their overall effect on the diagnostic process. Artifacts in 3D US may occur during image acquisition or reconstruction.
Artifacts are introduced during image acquisition when the relative position and angulation of some or all of the 2D images are cataloged incorrectly within the data array. This is usually caused by unplanned patient motion or transducer movement during acquisition. Involuntary patient motion (eg, cardiac motion, respiratory motion) (,Fig 7) also introduces artifacts. In general, the faster one can acquire the data, the less likely it is that these artifacts will occur. Generation of 3D images in real time with 2D arrays will eventually eliminate these artifacts (,9–,12). However, 2D arrays are still small and expensive; consequently, mechanical and tracked and untracked freehand data acquisition systems are likely to be a fact of life in 3D US for the next few years.
Although reconstruction algorithms are typically reliable and predictable, incorrect data assignments will occur if the machine is improperly calibrated (,Fig 8).
Suggestions for Optimal Imaging
The following are suggestions for optimizing image quality and minimizing artifacts.
1. It is always worthwhile to survey the anatomic area of interest prior to performing 2D US. This provides information regarding optimal scanning angles and mode of data acquisition and approximate size of the data set required. It also provides an opportunity to educate the patient about breath holding and the importance of remaining motionless and allows optimization of the 2D imaging parameters.
2. Machines should be properly calibrated according to manufacturer specifications.
3. For imaging with magnetic field tracking, nearby material that could distort the magnetic field should be removed.
4. The patient and the examiner should remain motionless during mechanical imaging. If possible, both the transducer assembly and the patient should be secured to the scanning table.
5. It is advisable to check each 3D image for motion artifacts immediately after imaging.
6. Use of copious amounts of warm gel is recommended to help minimize artifacts.
7. It is usually worthwhile to fine-tune the display during 3D imaging.
Applications for 3D US
Fetal Imaging
Three-dimensional US has shown the most promise in obstetric imaging. Rendering techniques give a novel perspective on the fetal anatomy (,Figs 9, ,10) and have been reported to make anomalies easier to recognize. Three-dimensional US also improves maternal-fetal bonding and can help families better understand fetal abnormalities.
Gynecologic Imaging
Evaluation of the pelvic anatomy with 2D US is limited by the orientation of the US beam, which allows viewing in the sagittal and coronal planes only. Three-dimensional US allows volume data sets to be acquired with both transvaginal and transabdominal probes (,Figs 11,–,,,13).
Three-dimensional Power Doppler Imaging
Dynamic “movies” of 3D US data are made by stacking together multiple individual images and playing them back in a fast cine loop. This technique provides a heightened sense of realism and has proved useful in 3D US angiography (,Fig 14). The value of 3D power Doppler imaging is best demonstrated on computer screens. Many good examples can be found at www.cs.uwa.edu.au/∼bernard/3dus/research.html and www.irus.rri.on.ca/3d-ultrasound.
Prostate Imaging
Three-dimensional prostate US allows more accurate repeated measurement than does 2D imaging (,,,,Fig 15). This can be very useful when accurate volume assessments are required for dosimetry planning or for estimating prostate-specific antigen levels.
Breast Imaging
Conventional US is very useful for distinguishing breast cysts, but 3D US has the capacity to demonstrate lesion margins and topography (,,,Fig 16). This information can be invaluable in differentiating benign from malignant masses. The breast tissue must be immobilized during imaging to ensure accurate registration of the imaging sections.
Biopsy-related Imaging
Three-dimensional US can be very useful in determining the likelihood of malignancy and therefore the need for biopsy. In addition, 3D US can help facilitate needle localization and guidance during biopsy (,,,,Fig 17).
Conclusions
Three-dimensional US is rapidly gaining popularity as it moves out of the research environment and into the clinical setting. This modality has some distinct advantages over conventional US and, with recent advances in computer technology and display techniques, promises to play an increasingly important role in medicine.
Figure 1a. Tracked freehand 3D scanning. (a) Acoustically tracked 3D scanning. Diagram shows a triangular device consisting of three sound emitters mounted on a US transducer. As the operator moves the transducer, the sound-emitting devices are activated and begin to emit pulses that are detected by three microphones positioned in different locations above the patient. By measuring the time delay between sound emission and detection, the position and angulation of the transducer are continuously monitored. (b) Articulated-arm-tracked 3D scanning. Diagram shows a transducer mounted on a mechanical arm with multiple movable joints. Potentiometers at the joints are used to measure their rotation, allowing continuous monitoring of the position and angulation of the transducer. (c, d) Magnetic field-tracked 3D scanning. (c) Diagram shows a small electromagnetic device attached to a transducer. The adjacent box detector registers nearby changes in the electromagnetic field. The exact position of the transducer is computed based on the changes in electromagnetic signal detected by the box detector. (d) Photograph shows an abdominal scan being performed with a magnetic tracking device, which is attached to the transducer (T) with a black plastic cover (arrow). The box detector (D) is attached to the scanning table.
Figure 1b. Tracked freehand 3D scanning. (a) Acoustically tracked 3D scanning. Diagram shows a triangular device consisting of three sound emitters mounted on a US transducer. As the operator moves the transducer, the sound-emitting devices are activated and begin to emit pulses that are detected by three microphones positioned in different locations above the patient. By measuring the time delay between sound emission and detection, the position and angulation of the transducer are continuously monitored. (b) Articulated-arm-tracked 3D scanning. Diagram shows a transducer mounted on a mechanical arm with multiple movable joints. Potentiometers at the joints are used to measure their rotation, allowing continuous monitoring of the position and angulation of the transducer. (c, d) Magnetic field-tracked 3D scanning. (c) Diagram shows a small electromagnetic device attached to a transducer. The adjacent box detector registers nearby changes in the electromagnetic field. The exact position of the transducer is computed based on the changes in electromagnetic signal detected by the box detector. (d) Photograph shows an abdominal scan being performed with a magnetic tracking device, which is attached to the transducer (T) with a black plastic cover (arrow). The box detector (D) is attached to the scanning table.
Figure 1c. Tracked freehand 3D scanning. (a) Acoustically tracked 3D scanning. Diagram shows a triangular device consisting of three sound emitters mounted on a US transducer. As the operator moves the transducer, the sound-emitting devices are activated and begin to emit pulses that are detected by three microphones positioned in different locations above the patient. By measuring the time delay between sound emission and detection, the position and angulation of the transducer are continuously monitored. (b) Articulated-arm-tracked 3D scanning. Diagram shows a transducer mounted on a mechanical arm with multiple movable joints. Potentiometers at the joints are used to measure their rotation, allowing continuous monitoring of the position and angulation of the transducer. (c, d) Magnetic field-tracked 3D scanning. (c) Diagram shows a small electromagnetic device attached to a transducer. The adjacent box detector registers nearby changes in the electromagnetic field. The exact position of the transducer is computed based on the changes in electromagnetic signal detected by the box detector. (d) Photograph shows an abdominal scan being performed with a magnetic tracking device, which is attached to the transducer (T) with a black plastic cover (arrow). The box detector (D) is attached to the scanning table.
Figure 1d. Tracked freehand 3D scanning. (a) Acoustically tracked 3D scanning. Diagram shows a triangular device consisting of three sound emitters mounted on a US transducer. As the operator moves the transducer, the sound-emitting devices are activated and begin to emit pulses that are detected by three microphones positioned in different locations above the patient. By measuring the time delay between sound emission and detection, the position and angulation of the transducer are continuously monitored. (b) Articulated-arm-tracked 3D scanning. Diagram shows a transducer mounted on a mechanical arm with multiple movable joints. Potentiometers at the joints are used to measure their rotation, allowing continuous monitoring of the position and angulation of the transducer. (c, d) Magnetic field-tracked 3D scanning. (c) Diagram shows a small electromagnetic device attached to a transducer. The adjacent box detector registers nearby changes in the electromagnetic field. The exact position of the transducer is computed based on the changes in electromagnetic signal detected by the box detector. (d) Photograph shows an abdominal scan being performed with a magnetic tracking device, which is attached to the transducer (T) with a black plastic cover (arrow). The box detector (D) is attached to the scanning table.
Figure 2a. Untracked freehand 3D scanning. (a) Diagram illustrates the mechanical movement of the transducer across the skin. (b) Diagram shows how the transducer can be tilted about a fixed point on the skin surface. (c) Diagram shows how the transducer can be rotated about its own axis.
Figure 2b. Untracked freehand 3D scanning. (a) Diagram illustrates the mechanical movement of the transducer across the skin. (b) Diagram shows how the transducer can be tilted about a fixed point on the skin surface. (c) Diagram shows how the transducer can be rotated about its own axis.
Figure 2c. Untracked freehand 3D scanning. (a) Diagram illustrates the mechanical movement of the transducer across the skin. (b) Diagram shows how the transducer can be tilted about a fixed point on the skin surface. (c) Diagram shows how the transducer can be rotated about its own axis.
Figure 3a. Mechanical 3D scanning. Photographs show a US transducer mounted in a plastic cradle (C) that is attached to a motorized device (M). The assembly is held completely still during acquisition. (a) Linear scanning. On activation, the motor moves the cradle and transducer down the tracks at a constant rate in a smooth, consistent manner (arrow). (b) Tilting scanning. On activation, the motor tilts the cradle and transducer at a constant rate in a smooth, consistent manner (arrow). (c) Rotational scanning. On activation, the motor rotates the cradle and transducer at a constant rate in a smooth, consistent manner (arrow).
Figure 3b. Mechanical 3D scanning. Photographs show a US transducer mounted in a plastic cradle (C) that is attached to a motorized device (M). The assembly is held completely still during acquisition. (a) Linear scanning. On activation, the motor moves the cradle and transducer down the tracks at a constant rate in a smooth, consistent manner (arrow). (b) Tilting scanning. On activation, the motor tilts the cradle and transducer at a constant rate in a smooth, consistent manner (arrow). (c) Rotational scanning. On activation, the motor rotates the cradle and transducer at a constant rate in a smooth, consistent manner (arrow).
Figure 3c. Mechanical 3D scanning. Photographs show a US transducer mounted in a plastic cradle (C) that is attached to a motorized device (M). The assembly is held completely still during acquisition. (a) Linear scanning. On activation, the motor moves the cradle and transducer down the tracks at a constant rate in a smooth, consistent manner (arrow). (b) Tilting scanning. On activation, the motor tilts the cradle and transducer at a constant rate in a smooth, consistent manner (arrow). (c) Rotational scanning. On activation, the motor rotates the cradle and transducer at a constant rate in a smooth, consistent manner (arrow).
Figure 4a. Two-dimensional array scanning. With this acquisition method, the transducer obtains true 3D data from an array of detectors. (a) Diagram illustrates a pyramid-shaped pulse originating from a square-faced transducer. (b) Diagram illustrates a cylindric pulse originating from a circular-faced transducer.
Figure 4b. Two-dimensional array scanning. With this acquisition method, the transducer obtains true 3D data from an array of detectors. (a) Diagram illustrates a pyramid-shaped pulse originating from a square-faced transducer. (b) Diagram illustrates a cylindric pulse originating from a circular-faced transducer.
Figure 5. Multiplanar reformatting with texture mapping. Three-dimensional color Doppler US image of the carotid artery shows the sharply jagged irregularity of the vessel wall and the color pattern caused by slight variations in the beat-to-beat movement of the artery. The image was acquired with cardiac gating, which improves the quality of the 3D image but increases total imaging time.
Figure 6. Volume rendering. Diagram shows how a ray interacts with a 3D volume image. The voxel values along each ray can be multiplied by selected factors and summed to produce different effects.
Figure 7. Respiratory motion artifact. Three-dimensional B-mode US image of the breast shows a cluster of small cysts (green arrow) but is severely degraded by undulations (yellow arrows) caused by the patient's rapid breathing pattern. This involuntary motion also caused smearing of the underlying image in a reconstructed plane (blue arrows). Details of the cluster of cysts are completely obscured, and the data had to be reacquired. A = acquired plane; B, C = reconstructed planes.
Figure 8. Artifact caused by incorrect calibration. Three-dimensional US image of the prostate gland acquired with rotational scanning with an end-fire transrectal US transducer shows a linear artifact in the middle of the gland (arrow). The artifact was caused by incorrect calibration of the scanning assembly. Note that no such artifact is seen in the near field. This artifact can be avoided by keeping the scanning assembly properly calibrated at all times.
Figure 9. Fetal imaging. Volume-rendered 3D US image clearly depicts the fingers of a fetus. 
Figure 10. Fetal imaging. Multiplanar reformatted 3D US image shows the umbilical vein, an umbilical artery, the inferior vena cava (IVC), and the fetal heart. These complex data can be displayed on a single image. Data acquisition time for this scan was less than 20 seconds.
Figure 11. Gynecologic imaging. Oblique-coronal endovaginal 3D US image demonstrates the uterine cornua (arrows) from a novel perspective. Endovaginal 3D US can be performed in less than 10 seconds, making it convenient for both patient and physician. Several authors have reported this modality to be useful in depicting a variety of clinical diseases, particularly congenital malformations of the uterus.
Figure 12. Gynecologic imaging. Multiplanar reformatted 3D US image shows first-trimester twins with areas of implantation bleeding (arrows) adjacent to the gestational sacs (s1, s2). Three-dimensional US allowed accurate measurement of the areas of bleeding as well as long-term follow-up.
Figure 13a. Gynecologic imaging. (a) Three-dimensional US image demonstrates enlarged ovarian follicles, whose sharp boundaries make volume measurement relatively easy. (b) On a 3D US image, one follicle has been delineated in green with the computer. Many assisted reproductive techniques cause ovarian hyperstimulation with resulting enlargement of the ovarian follicle. It is important to monitor follicular volume during a menstrual cycle to time fertilization accurately, but such monitoring is tedious with 2D US.
Figure 13b. Gynecologic imaging. (a) Three-dimensional US image demonstrates enlarged ovarian follicles, whose sharp boundaries make volume measurement relatively easy. (b) On a 3D US image, one follicle has been delineated in green with the computer. Many assisted reproductive techniques cause ovarian hyperstimulation with resulting enlargement of the ovarian follicle. It is important to monitor follicular volume during a menstrual cycle to time fertilization accurately, but such monitoring is tedious with 2D US.
Figure 14. Power Doppler imaging. Multiple volume-rendered images (A-D) from a power Doppler US angiography study demonstrate the normal spleen. Each image represents a frame from a cine loop.
Figure 15a. Prostate imaging. (a) Ejaculatory duct cyst. Coronal 3D US image of the prostate gland shows a midline ejaculatory duct cyst with a classic “teardrop” shape (arrowheads). The relationship of the cyst, the seminal vesicles (sv), and the verumontanum (yellow arrow) is clearly depicted. White arrow indicates pubic symphysis. (b, c) Right-sided base tumor with tumoral invasion. (b) Three-dimensional US image shows a hypoechoic tumor (arrows) invading the seminal vesicles (sv), a finding that indicates that the patient probably should not undergo surgery. rw = rectal wall. (c) Three-dimensional US image again shows the hypoechoic tumor (arrows) invading the seminal vesicle (sv). A large defect caused by transurethral resection of the prostate gland is also seen (arrowhead).
Figure 15b. Prostate imaging. (a) Ejaculatory duct cyst. Coronal 3D US image of the prostate gland shows a midline ejaculatory duct cyst with a classic “teardrop” shape (arrowheads). The relationship of the cyst, the seminal vesicles (sv), and the verumontanum (yellow arrow) is clearly depicted. White arrow indicates pubic symphysis. (b, c) Right-sided base tumor with tumoral invasion. (b) Three-dimensional US image shows a hypoechoic tumor (arrows) invading the seminal vesicles (sv), a finding that indicates that the patient probably should not undergo surgery. rw = rectal wall. (c) Three-dimensional US image again shows the hypoechoic tumor (arrows) invading the seminal vesicle (sv). A large defect caused by transurethral resection of the prostate gland is also seen (arrowhead).
Figure 15c. Prostate imaging. (a) Ejaculatory duct cyst. Coronal 3D US image of the prostate gland shows a midline ejaculatory duct cyst with a classic “teardrop” shape (arrowheads). The relationship of the cyst, the seminal vesicles (sv), and the verumontanum (yellow arrow) is clearly depicted. White arrow indicates pubic symphysis. (b, c) Right-sided base tumor with tumoral invasion. (b) Three-dimensional US image shows a hypoechoic tumor (arrows) invading the seminal vesicles (sv), a finding that indicates that the patient probably should not undergo surgery. rw = rectal wall. (c) Three-dimensional US image again shows the hypoechoic tumor (arrows) invading the seminal vesicle (sv). A large defect caused by transurethral resection of the prostate gland is also seen (arrowhead).
Figure 16a. Breast imaging. Multiplanar reformatted 3D US images of a portion of the nipple-areolar complex ) (a) and the nipple in profile (b) demonstrate duct ectasia and multiple simple cysts (green arrows). The oblique plane allows visualization of the serpiginous ducts (yellow arrow) from the cysts to the nipple (blue arrows) and facilitates understanding of the 3D imaging pattern of the disease.
Figure 16b. Breast imaging. Multiplanar reformatted 3D US images of a portion of the nipple-areolar complex ) (a) and the nipple in profile (b) demonstrate duct ectasia and multiple simple cysts (green arrows). The oblique plane allows visualization of the serpiginous ducts (yellow arrow) from the cysts to the nipple (blue arrows) and facilitates understanding of the 3D imaging pattern of the disease.
Figure 17a. Three-dimensional US-guided breast biopsy. A needle was advanced to the edge of the breast lesion, and 3D US was performed. (a) Three-dimensional US image obtained in the longitudinal (A) and transaxial (B) planes demonstrates the path of the needle. In the longitudinal plane, the needle (arrow) appears to reach the lesion (arrowheads), but the transaxial plane indicates that the needle is to the right of center. C = coronal plane. (b) On a 3D US image obtained in the transaxial (B) and coronal (C) planes, the needle (arrow) is high relative to the lesion (arrowheads). A = longitudinal plane. (c) Three-dimensional US image obtained in the longitudinal (A) and coronal (C) planes helps confirm that the needle (arrow) will miss the lesion (arrowheads). The needle was readjusted, and a satisfactory biopsy specimen was obtained.
Figure 17b. Three-dimensional US-guided breast biopsy. A needle was advanced to the edge of the breast lesion, and 3D US was performed. (a) Three-dimensional US image obtained in the longitudinal (A) and transaxial (B) planes demonstrates the path of the needle. In the longitudinal plane, the needle (arrow) appears to reach the lesion (arrowheads), but the transaxial plane indicates that the needle is to the right of center. C = coronal plane. (b) On a 3D US image obtained in the transaxial (B) and coronal (C) planes, the needle (arrow) is high relative to the lesion (arrowheads). A = longitudinal plane. (c) Three-dimensional US image obtained in the longitudinal (A) and coronal (C) planes helps confirm that the needle (arrow) will miss the lesion (arrowheads). The needle was readjusted, and a satisfactory biopsy specimen was obtained.
Figure 17c. Three-dimensional US-guided breast biopsy. A needle was advanced to the edge of the breast lesion, and 3D US was performed. (a) Three-dimensional US image obtained in the longitudinal (A) and transaxial (B) planes demonstrates the path of the needle. In the longitudinal plane, the needle (arrow) appears to reach the lesion (arrowheads), but the transaxial plane indicates that the needle is to the right of center. C = coronal plane. (b) On a 3D US image obtained in the transaxial (B) and coronal (C) planes, the needle (arrow) is high relative to the lesion (arrowheads). A = longitudinal plane. (c) Three-dimensional US image obtained in the longitudinal (A) and coronal (C) planes helps confirm that the needle (arrow) will miss the lesion (arrowheads). The needle was readjusted, and a satisfactory biopsy specimen was obtained.
∗∗. indicates multiple body systems. Abbreviations: 3D = three-dimensional 2D = two-dimensional
We thank Life Imaging Systems, London, Ontario, Canada for their assistance in producing the images that accompany this article.
References
- 1 Nelson TR, Pretorius DH. Three-dimensional ultrasound imaging. Ultrasound Med Biol 1998; 24:1243-1270. Crossref, Medline, Google Scholar
- 2 Pretorius DH, Nelson TR. 3-dimensional ultrasound in obstetrics: a review. Ultrasound Q 1998; 14:218-233. Crossref, Google Scholar
- 3 Downey DB, Fenster A. Three-dimensional ultrasound: a maturing technology. Ultrasound Q 1998; 14:25-40. Crossref, Google Scholar
- 4 Fenster A, Downey DB. 3-D ultrasound imaging: a review. IEEE Eng Med Biol 1996; 15:41-51. Google Scholar
- 5 Baba K, Jurkovic D, eds. Three-dimensional ultrasound in obstetrics and gynecology New York, NY: Parthenon, 1997. Google Scholar
- 6 Nelson TR, Downey DB, Pretorius DH, Fenster A. Three-dimensional ultrasound Philadelphia, Pa: Lippincott Williams & Wilkins, 1999. Google Scholar
- 7 Elliot TL, Downey DB, Tong S, McLean C, Fenster A. Accuracy of prostate volume measurements in vitro using three-dimensional ultrasound. Acad Radiol 1996; 3:401-406. Crossref, Medline, Google Scholar
- 8 Tong S, Cardinal HN, Downey DB, Fenster A. Analysis of linear, area, and volume distortion in 3D ultrasound imaging. Ultrasound Med Biol 1998; 24:355-373. Crossref, Medline, Google Scholar
- 9 Smith SW, Trahey GE, von Ramm OT. Two-dimensional arrays for medical ultrasound. Ultrason Imaging 1992; 14:213-233. Crossref, Medline, Google Scholar
- 10 Smith SW, Pavy HE, von Ramm OT. High speed ultrasound volumetric imaging system. I. Transducer design and beam steering. IEEE Trans Ultrason Ferromagn Freq Contr 1991; 38:100-108. Crossref, Google Scholar
- 11 von Ramm OT, Smith SW, Pavy HE. High speed ultrasound volumetric imaging system. II. Parallel processing and display. IEEE Trans Ultrason Ferromagn Freq Contr 1991; 38:109-115. Crossref, Google Scholar
- 12 Stetten GD, Ota T, Ohazama CJ, et al. Real-time 3D ultrasound: a new look at the heart. J Cardiovasc Diagn Proc 1998; 15:73-84. Google Scholar
