Multi-detector computed tomography (MDCT) systems may prove to be as revolutionary a development for clinical diagnostic imaging today, as the original application of single detector per slice CT was to the evaluation of head trauma in the early 1970s. The potential impact of the faster acquisition speeds, higher spatial resolution, and better image quality achieved by state-of-the-art MDCT technology includes new clinical applications, greater diagnostic accuracy, more efficient workflow, increased productivity, and better quality, cost-effective patient care. MDCT has the potential to change the way in which radiologists and clinicians interact with and navigate through the increasing amounts of image data, jettisoning diagnostic radiology from the two-dimensional slice realm into the third dimension. The acronym MDCT may become a misnomer as traditional tomographic data acquisition, analysis, and display evolve toward computerized volumetric imaging.
CT HISTORICAL PERSPECTIVE
Computed tomography, one of the medical breakthroughs made possible by the invention of the computer, was introduced nearly four decades ago as a tomographic imaging technique that generates cross-sectional maps of the x-ray linear attenuation of tissues in the axial plane of the body. In 1972, the first CT scanner (EMI Mark1) was used in clinical practice to noninvasively generate images of internal structures of the head. It produced images with an 80 by 80 matrix of 3 mm pixel resolution and required approximately 4.5 minutes of scan time and 1.5 minutes of reconstruction time per pair of slices. With today’s technology, image matrix size is 512 by 512 at a spatial resolution of 0.5 mm, scan times are measured in fractions of a second, and image data reconstruction is performed in near real time.
The first-generation scanners used a translate-rotate paradigm with a pencil beam or parallel ray geometry. A pair of sodium iodide (NaI) detectors was used to measure the transmission of x-rays through the patient for the two different slices, and each detector was physically moved or translated across the patient. Beginning at a given angle, the x-ray tube and detector apparatus linearly translated across the field of view, generating approximately 160 data points per projection. The whole tube-detector system was then rotated one degree and another projection obtained; 180 projections were obtained over a 180-degree rotation at 160 points per projection for a total of 28,800 rays.
The early detector systems could not accommodate the large change in x-ray flux that occurred as the system translated and measured rays from the thickest part of the head to the adjacent area, so the patient’s head was fitted into a flexible membrane surrounded by a water bath, the effect of which was to bolus the x-ray intensity of the beam outside the patient’s head to be similar in intensity to that inside the head. The NaI detector had a slow signal decay limiting the frequency with which the next measurements could be taken, thus contributing to the long scan time. One advantage of the first-generation CT scanners was the very efficient scatter reduction of the pencil beam geometry.
Second-generation scanners also used the translate-rotate technology but with multiple detectors and a narrow fan beam geometry. The incorporation of a linear array of 30 detectors and a divergent fan beam of x-rays allowed larger rotational increments and shorter scan times of 18 seconds per slice generating a single image in approximately 1 minute.
Third-generation CT scanners introduced in 1975 used a wide rotating fan beam geometry coupled with a large array of more than 800 detectors. This combination formed an arc large enough to allow the x-ray beam to interrogate the entire patient. The geometric relationship between the x-ray tube and the detectors did not change as the apparatus rotated 360 degrees around the patient. This rotate-rotate paradigm of the mechanically joined x-ray tube and detector array eliminated the time-consuming translational motion required by the first- and second-generation scanners and allowed for faster scanning. Early third-generation scanners could deliver a slice in less than 5 seconds and newer systems had 0.5 second scan times. Due to the substantial increase in the number of detector elements, third-generation CT scanners were more expensive than their predecessors and suffered from the significant problem of ring artifacts if one or more of the detectors were faulty or improperly calibrated.
In the rotate-rotate geometry, the data from each detector correspond to a ring in the reconstructed image. The detectors toward the center of the array provide the data for the smaller-diameter, more centrally located ring portion of the image, and the more peripheral detectors contribute to the larger-diameter ring portions. Perfect image reconstruction requires all of the detectors to be in perfect balance with each other. Because each detector and its associated electronics can have a certain amount of drift, causing signal levels to shift over time, the third-generation scanners could exhibit ring artifacts.
While modern detectors and more sophisticated calibration software can render third-generation CT scanners free of ring artifacts, fourth-generation scanners were designed in the late 1970s specifically to overcome them. Fourth-generation CT systems used a rotate-fixed paradigm with a rotating x-ray tube and a fixed ring of up to 4,800 detectors in the gantry. The divergent beam emerging from the x-ray tube forms a fan-shape with each detector acting as the vertex of the fan. (Compare this to the third-generation source-detector apparatus in which the x-ray tube acts as the vertex of the fan.) In fourth-generation scanners, the reference beam is measured by the same detector that is used for the transmission measurement, thus normalizing the data and eliminating ring artifacts.
Fifth-generation scanners, also known as electron-beam and sometimes called cine-CT devices, use an electron gun that deflects and focuses a fast-moving beam of electrons along a 210 degree arc of a large-diameter tungsten target ring in the gantry. The x-ray beam produced is collimated to traverse the patient and strike a detector ring. In this fixed-fixed geometry, there are no moving parts of the gantry, allowing images to be obtained in as little as 50 milliseconds with minimal motion artifacts. Fast-frame-rate CT movies of the beating heart can be acquired, making these systems popular with cardiologists. Electron-beam CT may also be useful for imaging patients who are unable to cooperate for routine examinations requiring a breath hold, such as pediatric and trauma patients.
Sixth-generation helical CT scanners use either third- or fourth-generation geometries with incorporated slip ring technology. A slip ring, introduced in the early 1990s, is a circular contact that allows the gantry to rotate continuously throughout the patient examination untethered by wiring to the scanner electronics. The use of slip ring technology makes it possible to achieve greater gantry rotational velocities than systems not using a slip ring, allowing shorter scan times. In helical CT acquisitions, the patient is moved along the horizontal axis as the x-ray tube rotates around the patient, such that the central ray of the x-ray beam follows a helical path during the scan. Image reconstruction is obtained by interpolating projection data at selected locations along the patient axis.
By acquiring data while the table is moving, the total scan time can be reduced. An entire abdomen, for example, can be scanned in approximately 30 seconds. The volume of intravenous contrast required can be reduced, and patient throughput can be increased. The ability to cover a large volume in a single breathhold can avoid misregistration due to inconsistent inspirations. Image reformatting into the coronal, sagittal, or oblique plane is improved with helical CT, and images can be reconstructed at any level and in any increment. A key limitation, however, is that the slice thickness is determined by the collimation used.
MDCT TECHNICAL ADVANCES
Seventh-generation multiple detector array CT scanners were introduced in 1992. They improve upon the third-generation geometry and sixth-generation helical scanning by using a wider portion of the x-ray fan beam during data acquisition. Collimator spacing is wider so that more of the x-rays that are produced by the tube are used in creating the image data. With conventional single-detector array scanners, opening up the collimator increases slice thickness, which improves x-ray beam utilization but reduces the spatial resolution in the slice thickness or z-axis dimension. Slice thickness in MDCT is determined by the detector size rather than by the collimation width as in conventional CT, so spatial resolution can be increased and isotropic voxels (with equal resolution in the x-, y-, and z-dimensions) can be acquired.
For example, an MDCT scanner with four contiguous parallel rows of 5 mm wide detectors and 20 mm collimator spacing can be used to simultaneously record four adjacent image slices in a single rotation of the x-ray tube, making better use of tube output. For the same technique factors (kV and mAs), the number of x-rays being detected is four times that of a single detector array with 5 mm collimation. Furthermore, the data set from the four 5 mm multiple detector arrays can be used to produce true isotropic 5 mm slices, or data from adjacent arrays can be combined to yield 10, 15 or 20 mm slices, all from the same acquisition. The use of isotropic voxels for image processing improves the display quality of multiplanar reformations (coronal, sagittal, or oblique images generated from the original axial image data), and 3D surface and volume rendering give more realistic representations.
Current multiple detector array scanners make use of solid-state detectors composed of a scintillator tightly coupled to a photodetector. Multiple detector arrays are a set of several linear detector arrays tightly abutted. Slice width is determined by grouping one or more detector units together. Clinical MDCT scanners include four-slice systems introduced in 1998 and 16-slice systems. The medical imaging industry has demonstrated 32-, 40- and the recently announced 64-slice capability. One company is working with a 256-slice prototype, which is said to be able to generate “real-time” 3D images. Increasing the number of active detector arrays may continue, culminating in area detectors being used for CT.
The flexibility of acquisition protocols and increased efficiency and speed of MDCT scanners achieve better patient imaging in a variety of ways. Reduced scanning time via the extremely rapid acquisition of conventional thick slices, or increased spatial resolution generating many high-resolution thin slices, can be achieved.
Many of the advantages of MDCT from a clinical standpoint arise from the speed and increased spatial resolution of the scanners, which in turn can improve the overall image quality. MDCT offers improved temporal resolution as faster scanning results in fewer motion artifacts. Breath-holding times can be reduced and patient comfort improved. Patient throughput may be increased, raising departmental productivity. Increased spatial resolution along the z-axis with thinner sections reduces partial volume artifacts and increases diagnostic accuracy. Because the scanning is done more quickly, less contrast material may be needed and patients may receive less radiation exposure. More efficient x-ray tube use and decreased image noise are possible with MDCT. Greater anatomic coverage can be achieved leading to more clinical applications.
The use of MDCT for neurological (brain perfusion, stroke assessment), oncologic, and thoracic imaging appears promising, but perhaps the biggest new clinical area is in cardiovascular imaging. Angiographic examinations as well as cardiac function measurements (coronary calcium scoring, ejection fraction evaluation, wall motion) are possible. Because scanning with MDCT is done more quickly, contrast material can be administered at a faster rate, improving the conspicuity of arteries, veins, and pathological conditions of blood flow.
High-resolution musculoskeletal, presurgical, and trauma imaging also appear promising. Positioning of the body in the gantry becomes less critical with MDCT because any plane can be reformatted from the acquired volume of data. Three-dimensional reconstructions with shaded-surface display or volume rendering can be performed to visualize bone, joints, and appliances.
A key advantage of MDCT speed is its ability to image a large volume (ie, chest-abdomen–pelvis) very quickly. The combination of isotropic volumetric imaging and high resolution enables high-quality 3D display of even curved structures such as the pancreatic duct or biliary tree. In solid organs, there is improved lesion detection due to the reduced motion artifact and thinner sections. MDCT provides the data for more realistic virtual colonoscopy and 3D surgical planning models. CT fluoroscopy and interventional guidance is another potential area for MDCT. Likely, many more applications will be explored.
ECONOMICS AND WORKFLOW
Some experts believe that the clinical merits of MDCT alone will sell the technology, while others believe that single-slice CT will remain a useful component of clinical care for several years, with hospitals waiting to buy MDCT during the normal replacement cycle. Still others may believe that the fast acquisition of fine resolution data will speed and improve diagnostic capabilities, translating into cost savings so that purchasing sooner rather than later is a good strategy. Some hospitals view the purchase of MDCT as a step in capturing a larger share of the patient market, with a portion of that arising from the ability to image pediatric and adult patients unable to hold their breath long enough to be scanned on the slower devices.
Others see MDCT as a potential for them to become a market leader by offering new CT services, drawing patients away from other modalities and invasive diagnostic procedures. For example, emergency medicine diagnostic protocols in the past may have included a stepwise approach to imaging beginning with conventional projection radiography, followed by CT and then perhaps MRI. Today MDCT may provide sufficient diagnostic information as the first and perhaps only step. It may be possible to generate additional revenue with MDCT because examinations are more likely to include reimbursement-eligible 3D reconstructions than studies from single-slice CT. There may also be cost savings in contrast media, as the shortened scan time for MDCT reduces the amount required.
Improved productivity is often one of the first benefits of new technology and MDCT is no exception. In theory, increased throughput of 30% to 40% may not be unreasonable to expect, as the machine is no longer the bottleneck. Studies have shown increased productivity and throughput in spite of the fact that multislice studies often involve more complicated imaging protocols. In practice, however, workflow changes may be necessary, and patient mix and the time required to get patients in and out of the scanner may become the rate-limiting step.
Direct acquisition and imaging costs of MDCT are not significantly different from those of single-slice technology. That is, the price of a multi-detector CT today is about the same as the price of a single-detector scanner purchased new 15 years ago. Personnel requirements may be less for some applications of MDCT, providing a cost benefit (ie, conventional angiography requires direct nursing and attending radiologist involvement while CT angiography does not). While MDCT offers economic benefits, there are some associated costs that come with its implementation. Maintenance costs and service contracts may run slightly higher than on less complex devices, and replacement costs of components such as the x-ray tube can be yearly and significant.
CHALLENGES AND FUTURES
It is likely that many health care enterprises will operate at least initially in a mixed environment with single- and multi-slice devices. During the transition to an MDCT-only environment, good imaging triage is important. It may be best to continue using conventional CT for the more straightforward cases that can be adequately served by single-slice scanners, while migrating to the newer devices those cases best suited to the strengths of MDCT, such as CT angiography.
MDCT contributes significantly to an escalating problem in diagnostic imaging: how to manipulate, transmit, store, access, display, analyze, and navigate through the massive amounts of data generated by new imaging modalities and procedures. Radiology departments should not even think about purchasing MDCT without a picture archiving and communications system (PACS), as the data files are large even for the digital world. A typical helical CT study generates 80 to 100, 512 by 512 slices. A four-slice CT scanner generates 150 to 500 slices; an eight-slice device generates twice as many; and a 16-slice scanner increases the output by a factor of four over the four-slice machines, generating 500 to 2,000 slices or between 800 and 1,000 MB (or 1 GB) of data per examination.
At the present time, most institutions store only the raw data generated by the MDCT temporarily and archive just the reconstructed data permanently. With this strategy, some of the capability to manipulate the image data will be lost. The reconstruction algorithm can no longer be changed nor the slice thickness decreased without the original data.
Not only is data storage an issue, but image viewing and navigation may require some workflow modifications. Scrolling through the axial slices of an MDCT examination is impractical. Innovative image interpretation paradigms are needed to maximize the utility of this modality in the future. Since MDCT data consist of isotropic voxels, the spatial and contrast resolution is preserved in all planes, so images can be viewed in any plane without distortion. Conversion to 3D volume displays and virtual reality fly-through representations may be a solution for the future.
Katherine P. Andriole, PhD, is assistant medical director Medical Imaging Information Technology Programs and director, Imaging Informatics Research, at the Center for Evidence-Based Imaging (CEBI), Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School.