Whether the modality is CT, MR, or ultrasonography, the move is to three-dimensional (3D) imaging. For this purpose, various mathematical formulas are used to determine how much of the data should be displayed and how those data should be weighted (see box, page 30). The display method chosen depends on the purpose of the examination. Three-dimensional imaging is improving diagnostic capabilities, making new imaging studies possible, and aiding the advance of minimally invasive surgery.
APPROACHES TO RECONSTRUCTION
Shaded surface display (SSD; also called surface rendering) was the first method to be introduced clinically and is widely available in equipment software. For simple thresholding, voxels with a certain density (eg, a particular range of Hounsfield units in a CT scan) are selected to create apparent surfaces. In a more sophisticated algorithm, called marching cubes, surface contours are modeled as overlapping polygons. With either technique, anatomy closer to the viewer appears brighter. The SSD method is fast and flexible, but no more than 10% of the data are used, and nothing below the surface is visible. Also, SSD is not helpful for anatomic regions in which there are no well-differentiated natural surfaces, and the images often do not clearly distinguish objects of similar density. The displays can lead to misjudgments of the severity of stenoses.
For another traditional method, maximum intensity projection (MIP), a series of parallel rays is projected through the dataset, and the voxels with the highest values are used to create the image. Alternatively, voxels with the lowest values can be used (minimum intensity projection). In a more sophisticated method, called sliding thin-slab reconstruction, a series of overlapping images is captured in slices whose thickness (eg, 5 to 10 mm) is defined by the radiologist. As with SSD, no more than 10% of the data are used for MIP image creation. Different software programs may create somewhat different images from the same dataset. The MIP method has a number of drawbacks. For example, there usually are no depth cues, and volume averaging may create artifacts. Also, it may not be possible to display an object of interest if a nearby object has equal or greater density, and spatial relationships tend to be poorly depicted, especially in areas of complex anatomy. Many of these problems can be overcome, but such workarounds require time to implement.
Volume rendering (VR) has been available since the mid 1980s, but only recently has sufficient computer power been available to make it practical for clinical use.? This technique (also called multiplanar volume rendering or reconstruction [MPVR] and volumetric visualization and analysis [VVA]) makes use of the entire dataset, creating a cube that can be sliced and displayed in various ways depending on the parameters defined by the user. The technology permits superficial structures to be peeled away and enables the spectacular “fly-throughs” of virtual endoscopy. The current generation of CT and MR scanners includes software for VR. (For a discussion of the technical aspects of VR, see the review by Calhoun et al. 1 )
SOME USES OF 3D CT
Spiral CT, with its acquisition of anatomic volumes, was a breakthrough technology for 3D CT, and the capabilities have been further expanded by the introduction of multidetector array scanners. Among the many applications are thoracic scans, where 3D reconstruction may avoid pulmonary angiography or bronchoscopy. 2 Planning tumor resection, 3 determining the extent of bronchiectasis, 4 and assessing indeterminate nodules identified by lung cancer screening are other current uses. 5,6 Even when the 3D images are not the principal diagnostic tool, they can increase confidence in the correctness of a diagnosis and facilitate communication with surgeons and referring physicians. 4
At New Hanover Regional Medical Center, Wilmington, NC, 3D angiography on a multidetector array scanner is used for neurovascular studies, as a screening test prior to peripheral vascular interventions, and for preprocedural and postprocedural evaluation of aortic endografts. Three-dimensional CT angiography also is usually the method of choice for the diagnosis of pulmonary emboli.
Even with the processing capabilities of the multidetector scanner, 3D images are created on a workstation, according to Allen J. Meglin, MD, of Delaney Radiologists, who is based at the Medical Center. “Postprocessing takes 5 to 10 minutes per patient, so if you do it on your scanner, your throughput will be terrible,” Meglin notes. “With a workstation, by the time we have finished postprocessing, we have scanned two more patients.”
At the University of Illinois, 3D electron-beam tomography (EBT) is used for coronary angiography. “We can easily determine whether bypass grafts are open, diseased, or closed,” reports Vladimir Jelnin, MD, senior research specialist. “We also use it for imaging stents. The scan takes about 20 minutes, and we can complete the 3D reconstruction on the workstation in 2 or 3 minutes. This approach is reducing the number of patients who go to the catheterization laboratory, as well as the number who have negative studies once they get there.”
Three-dimensional EBT angiography is valuable in the evaluation of congenital heart disease. Even in young children, it often is possible to perform the brief study with sedation rather than general anesthesia. “With a small injection of contrast, it takes only a few minutes to get good images, and we can measure the size of septal defects, and determine the extent of stenosis of the pulmonary arteries,” Jelnin says. “In a few cases, we scanned patients who had spent a couple of hours in the catheterization laboratory without clear definition of their defects. The patient was then taken back to the laboratory, and defect closure was finished in 5 minutes.”
Numerous other applications of 3D CT have been described. In patients with head and neck cancer, MIP images demonstrate the patency of blood vessels and any compression or displacement by tumor, and SSD and multiplanar reconstruction help identify and characterize lymph nodes and the relations of the tumor to surrounding structures. 7 Three-dimensional CT also is valuable in examining donors for living related liver transplants 8 and for planning percutaneous kidney stone removal in the morbidly obese. 9 A study at the National Kochi Hospital in Japan suggested that 3D spiral CT might obviate endoscopic retrograde cholangiography, 10 and multidetector array CT with 3D reconstructions is being used for 3-minute examinations of patients with multiple trauma. 11
SOME USES OF 3D MR
The utility of 3D MR angiography has been demonstrated repeatedly. Increasingly, MIP reconstructions are being supplanted by the more accurate VR technique. 12,13 The popularity of MR cholangiography for the diagnosis of choledocholithiasis continues to grow. Here, too, accuracy may be improved by the use of VR. 14 Determination of the structure of plaque, with the goal of identifying vulnerable lesions, 15 detection of intraductal spread of breast cancer, 16 measurement of intestinal transit time, 17 identification of lumbar disc herniation, 18 determination of the cause of sensorimotor hearing loss or vertigo, 19 detection of leakage of silicone from breast implants,20 virtual colography, 21 and confirmation of stress incontinence 22 have been described in the past few years.
The Surgical Planning Laboratory in the MRI Division of the Department of Radiology at Brigham and Women’s Hospital in Boston is developing methods for high-speed registration of 3D images acquired with different modalities for surgical guidance. Images are updated during surgery to reflect changes in the tissue.
Ron Kikinis, MD, director of the Surgical Planning Laboratory, describes some of the recent work. “We have developed software called 3D Slicer, which is available for research purposes ( www.slicer.org ). One of our operating rooms is equipped with an open MRI magnet in which our neurosurgeons have done close to 800 operations with the aid of 3D imaging. There also have been about 200 prostate brachytherapy implants and lesser numbers of other procedures such as lumpectomy, spine surgery, and virtual endoscopy.”
The modality employed is not the critical issue in imaging for surgical guidance.
“Whether you use ultrasound, CT, or MR, you need 3D postprocessing to merge the images,” Kikinis points out. “One challenge is to develop algorithms that are fast enough. A commercial software product must be able to run in almost any radiology department, so you have to make some assumptions about personnel training, which typically does not include image processing, and the type of computer available. We do not have those limitations, and by the time our algorithms are ready to be used clinically, the necessary computers will have become sufficiently inexpensive to make them practical. We can do image registration in less than a minute with a finite element approach on our server, which has 24 CPUs (computer processing units) and 24 GB of RAM. I predict that within a year, this feature will be available on high-end commercial products.”
USES OF 3D ULTRASONOGRAPHY
Ultrasound data are more difficult than CT and MR data to process into three dimensions, but 3D sonograms have been possible for 15 years. 23 However, it was slow to catch on in the United States. One reason was that sonographers were too busy to master the new equipment. Just as important was the poor quality of the images. Thus, it was customary to perform a 2D study with a high-end machine and then bring in a 3D scanner, a time-consuming step that reduced the number of patients who could be examined in a day.
Driven by clinical demand, especially from cardiologists, manufacturers improved image quality and developed scanners capable of both high-quality 2D and 3D imaging in close to real or real time. Other manufacturers created registration devices, with which 3D imaging can be done on any high-end scanner. Three-dimensional sonography is likely to become the standard for some applications, notably cardiology, and has demonstrated its value for purposes such as measuring carotid artery stenosis, 24 abdominal examination, and biopsy guidance. 25
“The 3D images probably do not make the diagnosis in most cases, but they make the diagnosis more reliable,” stresses Michael L. Manco-Johnson, MD, chairman of the Department of Radiology at the University of Colorado Health Science Center. “Those images also contribute to our ability to determine exactly where a lesion is, an invaluable adjunct to many interventional procedures such as TIPS (transjugular intrahepatic portosystemic shunt).”
The volume-acquisition capabilities of the 3D scanners have great utility under some circumstances. “Infants in the neonatal ICU are very unstable, and the shorter the time they are exposed the better,” Manco-Johnson adds. “In general, it takes 15 to 20 minutes to do a standard brain scan, because the sonographer has to go through every millimeter of the brain in a sequential way, look at the images, and decide when to record the image. This examination can be difficult because of the bright lights. With a 3D machine, you put the transducer into the Isolette, take three sweeps, each of which requires 5 seconds, and within 2 minutes, you have captured the whole brain. Then you can look at the coronal, axial, and sagittal views on the workstation.”
Similar acquisition of anatomic volumes might enable rural hospitals to provide new services to patients if prices come down sufficiently. A properly trained sonographer could acquire a volume of the heart or other anatomic site with a 3D scanner or with the aid of the localizing device and send the data over the Internet or a T1 line or to a workstation at a medical center where a specialist could walk through the volume and make the diagnosis.
There have been arguments that the sheer bulk of data produced by newer imaging equipment will drive even the most reluctant to 3D imaging. Others disagree, but the growing capabilities seem likely to make even the smaller hospitals consider the utility of imaging in three (or four!) dimensions.
Judith Gunn Bronson, MS, is a contributing writer for Decisions in Axis Imaging News.
- Calhoun PS, Kuszyk BS, Heath DG, Carley JC, Fishman EK. Three-dimensional volume rendering of spiral CT data: theory and method. Radiographics. 1999;19:745-764.
- Johnson PT, Fishman EK, Duckwall JR, Calhoun PS, Heath DG. Interactive three-dimensional volume rendering of spiral CT data: current applications in the thorax. Radiographics. 1998;18:165-187.
- Remy J, Remy-Jardin M, Artaud D, Fribourg M. Multiplanar and three-dimensional reconstruction techniques in CT: impact on chest diseases. Eur Radiol. 1998;8:335-351.
- Salvolini L, Bichi Secchi E, Costarelli L, DeNicola M. Clinical applications of 2D and 3D CT imaging of the airways: a review. Eur J Radiol. 2000;34:9-25.
- Gruden JF, Ouanounou S, Tigges S, Norris SD, Klausner TS. Incremental benefit of maximum-intensity-projection images on observer detection of small pulmonary nodules revealed by multidetector CT. AJR Am J Roentgenol. 2002;179:149-157.
- Ravenel JG, McAdams HP, Remy-Jardin M, Remy J. Multidimensional imaging of the thorax: practical applications. J Thorac Imaging. 2001;16:269-281.
- Franca C, Levin-Plotnick D, Sehgal V, Chen GT, Ramsey RG. Use of three-dimensional spiral computed tomography imaging for staging and surgical planning of head and neck cancer. J Digit Imaging. 2000;13(2 suppl 1):24-32.
- Bogetti JD, Herts BR, Sands MJ, Carroll JF, Vogt DP, Henderson JM. Accuracy and utility of 3-dimensional computed tomography in evaluating donors for adult living related liver transplants. Liver Transpl. 2001;7:687-692.
- Buchholz NP. Three-dimensional CT scan stone reconstruction for the planning of percutaneous surgery in a morbidly obese patient. Urol Int. 2000;65:46-48.
- Ishikawa M, Tagami Y, Toyota T, et al. Can three-dimensional helical CT cholangiography before laparoscopic cholecystectomy be a substitute study for endoscopic retrograde cholangiography? Surg Laparosc Endosc Percutan Tech. 2000;10:351-356.
- Ptak T, Rhea JT, Novelline RA. Experience with a continuous, single-pass whole body multidetector CT protocol for trauma: the three-minute multiple trauma CT scan. Emerg Radiol. 2001;8:250-256.
- Mallouhi A, Schocke M, Judmaier W, et al. 3D MR angiography of renal arteries: comparison of volume rendering and maximum intensity projection algorithms. Radiology. 2002;223:509-516.
- Baskaran V, Pereles FS, Nemcek AA Jr, et al. Gadolinium-enhanced 3D MR angiography of renal artery stenosis: a pilot comparison of maximum intensity projection, multiplanar reformatting, and 3D volume-rendering postprocessing algorithms. Acad Radiol. 2002;9:50-59.
- Kondo H, Kanematsu M, Shiratori Y, et al. MR cholangiography with volume rendering: receiver operating characteristic curve analysis in patients with choledocholithiasis. AJR Am J Roentgenol. 2001;176:1183-1189.
- Coombs BD, Rapp JH, Ursell PC, Reilly LM, Saloner D. Structure of plaque at carotid bifurcation: high-resolution MRI with histological correlation. Stroke. 2001;32:2516-2521.
- Nakahara H, Namba K, Wakamatsu H, et al. Extension of breast cancer: comparison of CT and MRI. Radiat Med. 2002;20:17-23.
- Patak MA, Weishaupt D, Frohlich JM, Debatin JF. Sequential fast 3D MRI following oral ingestion of Gd-DOTA: a new means to assess intestinal transit time. J Magn Reson Imaging. 1999;10:474-476.
- Taira G, Endo K, Ito K, Ichimaru K, Imakiire A, Miura Y. Diagnosis of lumbar disc herniation by three-dimensional MRI. J Orthop Sci. 1998;3:18-“26.
- Held P, Fellner C, Fellner F, Seitz J, Graf S, Strutz J. Correlation of 3D MRI and clinical findings in the patients with sensorimotor hearing loss and/or vertigo. Clin Imaging. 1998;22:309-322.
- Harms SE, Jensen RA, Meiches MD, Flamig DP, Evans WP. Silicone-suppressed 3D MRI of the breast using rotating delivery of off-resonance excitation. J Comput Assist Tomogr. 1995;19:394-399.
- Luboldt W, Debatin JF. Virtual endoscopic colonography based on 3D MRI. Abdom Imaging. 1998;23:568-572.
- Hoyte L, Schierlitz L, Zou K, Flesh G, Fielding JR. Two- and 3-dimensional MRI comparison of levator ani structure, volume, and integrity in women with stress incontinence and prolapse. Am J Obstet Gynecol. 2001;185:11-19.
- Lees W. Ultrasound imaging in three and four dimensions. Semin Ultrasound CT MR. 2001;22:85-105.
- Kozakova M, Morizzo C, Andreucetti F, Palchetti P, Parenti G, Palombo C. Quantification of extracranial carotid artery stenosis by ultrafast three-dimensional ultrasound. J Am Soc Echocardiogr. 2001;14:1203-1211
- Weismann CF, Forstner R, Prokop E, Rettenbacher T. Three-dimensional targeting: a new three-dimensional ultrasound technique to evaluate needle position during breast biopsy. Ultrasound Obstet Gynecol. 2000;16:359-364.