Elliot K. Fishman, MD

The first announcements came at the 2003 RSNA meeting, informing the radiology community that 64-slice CT would soon be available for clinical practice. Initial site installations by Siemens Medical Solutions and GE Healthcare began shortly after, in late spring and early summer of 2004. The introduction of this new technology came as a surprise to many, since 16-slice CT had been widely available only since 2002.

Radiologists began to wonder if this new generation of scanners would be truly revolutionary or simply evolutionary. Our early experiences have encouraged us to examine the directions we think 64-slice CT will take in the near term. Although all the major CT vendors-GE, Philips, Siemens, and Toshiba-have installed at least one 64-slice CT scanner, we have used only the Siemens Sensation 64 scanner, and our discussion of technical features will focus on that system. Our review of the clinical potential and applications of 64-slice CT technology, however, is not vendor-specific.

The introduction of 16-slice CT addressed many of the challenges and limitations present in quadslice CT scanning. Sixteen-slice CT provided for routine subsecond scan rotation times (usually around 0.5 sec), routine submillimeter slice acquisition parameters (usually 0.5 to 0.75 mm), and true isotropic data sets. Combining these scanning parameters with the ability to reduce long acquisition times and problems with tube heating meant that multislice CT had met most of the challenges of true CT volume imaging. Not surprisingly, clinical applications also evolved to take advantage of these technical advancements in three main areas:

- Classic applications such as evaluation of the liver, pancreas, and kidneys could be done with improved scan protocols, leading to better detection of disease, more accurate staging, and improved patient management.

- Vascular applications commonly referred to as CT angiography could be expanded now that true isotropic volume acquisitions were possible, overcoming the limitations of quadslice CT.

- New applications such as coronary CT angiography and peripheral runoff studies, although attempted at select institutions with quadslice CT, became practical routine applications.

Although 16-slice CT soon became the accepted gold standard for CT imaging, the ability to combine fast acquisition with properly timed injection rates and isotropic resolution still faced certain inherent limitations, especially with applications such as coronary angiography or functional imaging. Thus, researchers immediately began to look to the next generation of scanners, which could improve on experience with 16-slice CT.

In times of rapid changes in scanner technology, it is not surprising that each manufacturer regards its product as a superior one that will provide the user with the optimal scanning environment. Two of the leading CT scanning vendors for the past two decades, GE and Siemens, have introduced and installed 64-slice scanners for clinical use in the U.S. and, in Siemens' case, Europe and Japan. Although both machines are defined in the generic sense as 64-slice scanners, the vendors point out some important differences that they consider key differentiators.

Since the number of installations is small and the winner or winners cannot be determined, we asked both vendors to define the attributes of their product from a technical perspective. We will not try to compare claims of superiority, but hope to clarify the performance of each system and let the test of time show the advantages. As noted, these parameters are provided by the manufacturers and not the authors.

The Siemens Sensation 64 scanner, through its new z-sampling technology, delivers < 0.4-mm isotropic resolution for pitch values up to 1.5 in routine studies, even at isocenter, without spiral artifacts. This is achievable at a coverage speed of up to 87 mm/sec. This speed is calculated using the best resolution, which is 0.4 mm, and by setting pitch to 1.5 and the feed/second at 0.33-sec, rotation time = 1.5 x 19.2/0.33 = 87 mm/sec. The 19.2 mm is calculated by a feed per rotation of 32 x 0.6 mm. Although the Siemens 64-slice system has 32 detector elements in the center of the assembly, it has a 64-slice data acquisition system behind those detectors. The new z-sampling scheme and this detector assembly enable the simultaneous readout of two projections. The detector readout of two times 32 slices for every viewing angle enables full 64-slice acquisition. The scanner has a rotation speed of 0.33 sec, which provides unique protocol possibilities that will have an impact across a range of applications.

The GE LightSpeed VCT scanner has 64 rows of 0.625-mm detectors. At a pitch of just under 1 (63/64 = 0.984), with 40 mm of full-resolution coverage, it will scan 39.38 mm per rotation with the same or better z-axis resolution as the LightSpeed 16, 0.36 mm as measured by IMPACT. The GE detector elements are essentially square (0.625 x 0.6 mm) at isocenter, and the scanner is intrinsically isotropic. LightSpeed VCT will have an optional 0.35-sec rotation. GE also claims that in cardiac-gated mode, the scanner can do a 12 to 15-cm cardiac acquisition in five to six seconds. On the LightSpeed VCT, in high-resolution mode (64 x 0.625 mm), the anatomic coverage is 40 mm per rotation. At 1:1 pitch and 0.37-sec rotation, this corresponds to 108 mm/sec.

Now that we have obtained these amazing scan parameters, the question becomes what the impact will be in clinical practice. We all experienced the revolutionary changes of 16-slice CT over quadslice but wonder whether 64-slice CT will spark a similar revolution. Although we are very early in the game, the answer is a resounding yes, no, and maybe. The path of 16-slice CT offers insight into the possible path of 64-slice CT.

There is little doubt that 16-slice CT is an impressive piece of technology. Even though it was developed and introduced into clinical practice only two years ago, most institutions have seen a double-digit growth of body CT scanning because of its capabilities.

At Johns Hopkins, our body CT scan volume increased over 20% in the last fiscal year, but the 3D/CTA volumes increased more than 40%. These numbers are not surprising to anyone who has witnessed CT's transformation into the physical examination of the 21st century. CT has replaced a range of other studies and is rapidly becoming the primary noninvasive vascular imaging study. But we recognize that as good as 16-slice CT is, definite limitations persist. These include speed of acquisition, image resolution, and image noise and artifact for select applications. This is where 64 will become the de facto standard. Yet most institutions will likely use a combination of 16- and 64-slice scanners in their clinical practice.

No doubt, the most exciting application for 64-slice CT is coronary artery and cardiac imaging. With 16-slice CT, we began doing coronary CTA. But even with the use of beta blockers, the study success rate was adequate in only 80% of patients. The problems related to a combination of acquisition time and slice thickness, or what might be described as temporal and spatial resolution. With 64-slice CT, we are able to use cardiac gating and acquire a data set in as little as five to seven seconds, or a few heartbeats.

The ability to visualize the coronary arteries in unprecedented detail without motion artifact is the promise that 64-slice CT brings. Acquisition of 4D data provides studies of cardiac motion, cardiac function, and valve function. The use of fast scanning also allows sequential imaging of the myocardium to look for perfusion changes of early myocardial infarction. Many centers are beginning to develop protocols to look at acute coronary syndrome with 64-slice CT as the initial, and potentially only, exam. If successful, this approach will revolutionize care in the emergency room and lead to changes in the management algorithm for millions of ER patients every year.

Vascular imaging beyond the heart and coronary arteries will also benefit from 64-slice technology. Two specific applications are evaluation of the carotid artery and peripheral runoff studies. Both have faced limitations with prior technologies. Carotid imaging-when done from the aortic arch through the base of the skull or even through the circle of Willis-was often degraded by rapid venous contamination. Although the use of editing tools, clip planes, and volume rendering made these studies possible, 64-slice technology will allow an acquisition fast enough to avoid venous contamination.

In peripheral studies, two issues are the timing of the bolus transit and whether this should be done at the aorta or popliteal vessels. Both strategies have faced limitations. With the 64-slice scanner, the scan can be monitored until contrast appears in the lower legs, triggering acquisition. This will potentially limit venous contamination caused by scanning too late or inadequate enhancement due to scanning too early.

Similarly, a 64-slice system with higher spatial and temporal resolution will no doubt provide better data sets for routine imaging, ranging from evaluation of suspected liver metastases and staging of renal tumors to detection of pulmonary embolism. Yet 16-slice CT is so good that improvement in sensitivity and specificity might be limited if a traditional approach accompanies the shift to 64-slice imaging. We expect 64-slice CT to result in a new paradigm. We will move from purely structural imaging of normal anatomy and pathology to a more functional approach. Functional or physiologic imaging not only detects discrete lesions per se but also recognizes increases or decreases in vascular flow.

To date, most functional CT imaging has targeted the brain, with varying success. In some institutions, it is a critical part of stroke imaging. But organ enhancement often is a result of a range of factors that go far beyond the presence of a discrete lesion or vascular invasion by a mass. Rather, functional flow changes may occur on a microscopic level where no detectable tumor mass is seen. Early changes on a vascular level due to tumor may result in changes in perfusion that could suggest the presence of early disease. Previously, articles have suggested that changes in liver perfusion may be a good indicator of future metastatic disease. In addition, flow-related changes in a kidney or liver mass may help define the degree of angiogenesis in a tumor and guide surgical management. We also think that perfusion may be a way to look for early pulmonary embolism, taking advantage of both visualization of the pulmonary arteries and lung perfusion, thereby matching some of the attributes of traditional lung perfusion scanning with nuclear medicine studies.

With development of new agents for PET scanning or even with current agents, a combination of perfusion mapping may go a long way to support continuing understanding of tumor activity and recurrence patterns using PET.

The introduction of 64-slice CT could suggest that all of our limitations with CT have been solved, and, as Charles H. Duell, U.S. commissioner of patents in 1899, said, "Everything that can be invented has been invented." In fact, nothing is further from the truth. Sixty-four-slice CT is but one step leading to area detectors, with which a scan of the heart or liver might be done in less than a heartbeat. The era of 64-slice CT creates new issues, reminding us of a combination of Richard Nixon's "Solutions are not the answer" and Eric Sevareid's "The chief cause of problems is solutions."

Radiologists are already concerned about network transfers, archiving, and retrieving of 16-slice CT data sets, which run in the 400 to 1200-slice range for most CTAs. What will they do with two to three times that number of slices from 64-slice CT? That challenge must be addressed. In addition, workstations have been upgraded in recent years with special processing boards to interactively render 1000 or more slices. But can they handle 3000 to 5000-slice data sets? What about the load time for these data sets on a workstation?

Radiologists and referring physicians are now becoming comfortable with volume visualization as a primary reading and review mode. But they are not willing-nor should they be willing-to wait 10 to 15 minutes for a data set to load. Storage of data on central servers as well as retrieval of data will become enterprise-wide issues that require solutions. Success in solving the backbone issues will go a long way toward ensuring the success of 64-slice CT. Surmounting the challenges to efficient workflow is critical if 64-slice CT is to reach its full potential.

We are confident that these problems and others will be overcome through careful collaboration between industry and academia. This approach should include radiologists, technologists, referring physicians, computer scientists, and design specialists. New models will evolve, and the radiologist must lead, rather than react to, this change.

The advent of 64-slice CT will probably lead to the implementation of an imaging workflow that truly incorporates volume visualization into routine practice. It has taken almost 20 years to get to this point. We are anxious to see what the next two years will bring.

Dr. Fishman is a professor of radiology and director of diagnostic radiology and body CT, and Dr. Horton is an associate professor of radiology, both at Johns Hopkins Hospital in Baltimore; Dr. Anders is a radiology resident, and Dr. Achenbach is an assistant professor of radiolgy, both at the University of Erlangen in Erlangen, Germany.

Articles

Clinicians weigh 64-slice CT's revolutionary potential

The first announcements came at RSNA 2003, informing the radiology community that 64-slice CT would soon be available for clinical practice. Initial site installations by Siemens Medical Solutions and GE Healthcare began shortly after, in late spring and early summer of 2004. The introduction of this new technology came as a surprise to many, since 16-slice CT had been widely available only since 2002.

The Siemens Sensation 64 scanner, through its new z-sampling technology, delivers  

CT has always played a prominent role in the evaluation of musculoskeletal pathology. With the advent of spiral CT and, most recently, multidetector row CT (MDCT), the data sets available for image analysis and for postprocessing and display have unprecedented image resolution and detail. Concurrent with this advance is the development of postprocessing techniques, especially three-dimensional volume rendering.

The newest scanners provide increased options for protocol design and optimization of both data reconstruction techniques and postprocessing tools. The quality of a CT data set depends on many factors, from mAs to kVp to slice thickness. Routine CT imaging of bone has classically required thin-section imaging (3 mm or less) and close interscan reconstruction intervals (3 mm or less), especially when multiplanar reconstruction or 3-D imaging is needed.

With MDCT, the user has increased flexibility; instead of selecting the slice thickness prior to the study, the user chooses the scan collimation. On our Siemens Somatom Volume Zoom scanner, we use four detectors for data acquisition from each 500-msec scan rotation. The detectors typically chosen are either 1 mm or 2.5 mm, with the 1 mm now clearly the selection of choice in small-part imaging (e.g., wrist or hand). With the 1-mm detector a range of slice thickness can be obtained. For example, we can reconstruct data at slice thicknesses ranging from 1 to 8 mm. For the best detail and for the best 3-D reconstruction we will select 1-mm to 1.25-mm slice thickness and reconstruct the data at 1-mm intervals. The scan protocol typically includes 120 kVp, 150 mAs and a pitch of six to eight. On our scanner, pitch is defined as distance traveled per scan rotation/nominal slice thickness (e.g., 15 mm/sec travel with 2.5 mm collimators is a pitch of six). For smaller volumes, we can obtain a 0.5-mm slice thickness, although pitch is limited to two, which results in longer scan times for the volume. However, the advantage of 0.5-mm slice thickness is that the volume becomes an isotropic data set with the x, y, and z axes equal in size, as opposed to routine CT data sets, which are anisotropic. The advantages of an isotropic data set is that regardless of the plane in which the patient is scanned, reconstructed data in any plane are of equal resolution.

A past concern with spiral CT and skeletal imaging was that the image quality suffered on both the axial and 3-D display as higher pitches were used. This was due in part to the less than optimal scan reconstruction algorithms available and also to how the single spiral scans were obtained. With higher pitches, the slice profile might increase up to 27% (that is, a scan with a 5-mm collimation might actually be 6.5 mm) and the mAs would actually decrease on the higher pitches, resulting in poorer image quality. With MDCT, the slice thickness selected is the slice thickness achieved, without any blooming artifact. Also, the mAs are constant across the scan volume, regardless of the pitch selected.

Other routine choices in musculoskeletal CT imaging are key, and they depend on the following decisions:

 Is IV contrast needed for the study? In cases of pelvic trauma, a CT angiogram may be done concurrently with the musculoskeletal imaging exam; numerous studies have shown that CTA is highly accurate in this situation. In addition, the presence of muscle pathology is best seen following infusion of IV contrast, regardless of whether one is looking at abscess, tumor, or myositis. 

 Will 3-D imaging or MPR (multiplanar reconstruction) be needed? These require use of thin collimation and close interscan spacing. 

 What anatomic area will need to be scanned? Scan protocols depend on the volume to be scanned (e.g., tibial plateau versus the entire tibia). 

 What reconstruction algorithm will need to be selected? The answer depends on the clinical application: In cases where soft tissue or muscle imaging is the primary process, a standard soft-tissue algorithm is used, but if bone detail is needed then a high-resolution algorithm is necessary. Note also that the ideal reconstruction algorithm will depend on whether 3-D rendering needs to be done. Some high-resolution algorithms will create images with too much noise and yield 3-D images of suboptimal quality. 

One of the classic applications for CT and for 3-D imaging is the evaluation of musculoskeletal trauma. Even in the earliest days of CT, many articles pointed out the advantages of CT and 3-D imaging as both a diagnostic and clinical management tool. Patients commonly referred to CT were sent either because plain films were indeterminate as to fracture, or in cases of extensive fracture, where the true extent of injury needed to be determined for surgical or nonsurgical planning (Figure 1). The 3-D images in these cases changed the patient's management in up to 30% of cases.

In the last 20 years, while little has changed in referral patterns for many trauma applications, other changes have occurred. CT scanners have become more available, speeds have increased, CT scanners have been placed in or near emergency rooms, and staffs have inaugurated systems for rapid triage of patients, with CT expediting diagnoses in crowded emergency rooms. In the ER, new imaging algorithms for spinal trauma as the initial study are becoming common. (Figure 2).

The use of MDCT is ideal in all trauma applications. CT can limit the need for sedation and provide a single comprehensive exam for imaging the solid organs and the skeletal structures. This is especially true in pediatric patients, for whom MDCT virtually has eliminated sedation.

The high-quality data sets generated by MDCT are important in small-part imaging, as in the wrist (Figures 3 and 4).

The use of 0.5-mm slice thickness and isotropic data sets are ideal for this application. Supplemented by 3-D renderings, they can produce all of the necessary views for surgical decision-making. Isotropic data sets eliminate the need for two views and speed up the examination without compromising quality.

In the evaluation of pelvic trauma, MDCT is ideal for combining skeletal imaging with more complicated studies including CT angiography, where it looks for vascular injury and where CT cystograms can help to exclude bladder injury in patients with pelvic trauma. CT cystograms are as accurate as classic cystograms for detecting bladder injury.

The role of CT scanning has continued strong even with the continuing evolution of MRI. Following are some clinical scenarios in which CT is used:

 Inconsistencies arise between interpretation of plain radiographs and patient symptoms, or between bone scans and plain films. 

 The extent of tumor needs determination before choosing between surgery and radiation therapy. 

 A requirement arises to define the potential biopsy of a lesion, where biopsy can be done with CT guidance and, if necessary, CT fluoroscopy. (CT fluoroscopy allows biopsy of smaller lesions in more difficult locations and can also guide radio-frequency ablation of osteoid osteoma.) 

 One needs to define the presence and extent of disease that may be detected with bone scanning or PET scanning, with CT able to help determine whether increased uptake in a bone scan is Paget's disease or metastases in indeterminate cases. 

The use of MDCT with optimal scan protocols is routinely supplemented by 3-D reconstructions in these patients (Figure 5). We routinely combine volumetric 3-D reconstruction with multiplanar coronal and sagittal reconstruction. The use of MDCT data as a volume display rather than as a slice-based display is critical, especially when looking at joint surfaces and articular zones. With ever-larger datasets being generated, the 3-D display will become the dominant display in the future.

CT has always evaluated congenital hip disease. Murray and Crim, in a recent review on the radiologic imaging and treatment of developmental dysplasias of the hip, recommended using multidetector CT with thin-section imaging.

In addition to the virtual elimination of sedation, MDCT provides ideal data sets for 3-D reconstruction. Numerous authors have documented that CT and 3-D imaging are invaluable in both preoperative planning and postoperative follow-up in patients with developmental dysplasias of the hip. The more severe the deformity, the more valuable the reformation of the data set becomes. In younger patients, it is important to select the lowest scan parameters possible to minimize the radiation dose to the patient.

Similarly, in areas such as the chest wall, imaging can be valuable in detecting benign anatomic variations that can simulate disease. We have also found CT to be helpful in the evaluation of a patient's pre-pectus excavatum repair as well as in patients with failed pectus repairs (Figure 6). In the latter group of patients, Pretorius et al previously showed how management is affected by the use of 3-D CT.

The use of isotropic data sets with 0.5-mm collimation is ideal for evaluation of the painful foot where a tarsal coalition is one of the more common diagnoses. The ability to reconstruct data in any plane obviates a second set of scans and provides a volume display that can help detect and define bony or fibrous unions. It is not surprising that in cases where tarsal coalition is suspected that other diagnoses are made, based on the CT scan, to explain the patient's symptoms.

One of the most challenging clinical scenarios occurs in the patient who has had prior orthopedic surgery with the insertion of pins, plates, screws, or joint replacement. Classic axial CT may be limited due to artifact generated on axial CT scans caused by the metal implants. Image postprocessing can solve this problem. Some CT scanners provide postprocessing software that can reduce artifact or extend the CT scale to minimize the artifacts generated as a result of beam hardening.

A technique that we find especially valuable is the generation of multiplanar reconstruction and/or 3-D volume renderings, using thin collimation and narrow interscan spacing. Since artifact typically is random on each individual CT slice, the reconstruction of overlapping CT scans will limit the artifacts and result in diagnostic-quality CT images. Specific applications where we have found this valuable include assessment of fracture healing or suspected postoperative infection, evaluation of suspected failure of orthopedic hardware, and evaluation of patients with persistent pain with hardware in place. When using 3-D we have found the ideal: using color for metal (blue works best) against the white opacification usually selected for bone. Our experience with MDCT involves slice thicknesses ranging ideally from 0.5 to 1.25 mm, with a 0.5 to 1-mm reconstruction interval (Figure 7).

The role of CT in the evaluation of musculoskeletal pathology will continue to grow and evolve with the diffusion of MDCT and its evolution. New patient protocols will be developed, especially in the emergency room setting, where CT in many cases will be the initial and only study performed in patients with trauma. The use of CT in a volume display mode will also be ideal for rapid screening of the multisite trauma patient and will potentially become a cost-effective, time-effective study technique. The next generation of MDCT scanners will also allow routine use of isotropic data sets of 0.5-mm thickness. This is ideal for high-resolution imaging, which will be especially valuable in musculoskeletal imaging, especially for small-part CT scanning. Whether for trauma, infection, or oncologic applications, the role of CT is just beginning to be understood.

Dr. Fishman is a professor of radiology and oncology at Johns Hopkins University.

 Daffner R. Cervical radiography for trauma patients: a time effective technique? AJR 2000;175:1309-1311. 

 Donnelly LF. Use of three-dimensional reconstructed helical CT images in recognition and communication of chest wall anomalies in children. AJR 2001;177:441-445. 

 Fishman EK, Magid D, Robertson DD, et al. Reconstruction metallic hip implants: CT with multiplanar radiology. Radiology 1986:160:675-681. 

 Goldfarb CA, Yin Y, Gilula LA, et al. Wrist fractures: what the clinician wants to know. Radiology 2001;219:11-28. 

 Kuszyk BS, Ney DR, Fishman EK. The current state of the art in 3-D oncologic imaging: an overview. Int J Radiation Oncology Biol Phys 1995;33:1029-1039. 

 Kuszyk BS, Heath DG, Bliss DF, Fishman EK. Skeletal 3-D CT: Advantages of volume rendering over surface rendering. Skeletal Radiol 1996;25:207-214. 

 Link TM, Berning W, Scherf S, et al. CT of metal implants: reduction of artifacts using an extended CT scale technique. J Comput Assist Tomogr 2000;24(1):165-172 

 Murray KA, Crim JR. Radiographic imaging for treatment and follow-up of developmental dysplasia of the hip. Sem Ultra CT MRI 2001;22(4):306-340. 

 Ney DR, Drebin RA, Fishman EK, Magid D. Volumetric rendering of computed tomographic data: principles and techniques. IEEE Comput Graphic Appl 1990;10:24-32. 

 Pretorius ES, Haller JA, Fishman EK. Spiral CT with 3-D reconstruction in children requiring reoperation for failure of chest wall growth after pectus excavatum surgery: preliminary observations. Clin Imaging 1998;22:108-116. 

Elliot K. Fishman, MD

Articles

Clinicians weigh 64-slice CT's revolutionary potential

Clinicians weigh 64-slice CT's revolutionary potential

3-D renderings find role in musculoskeletal pathology studies