The emergence of 4-row multidetector CT (MDCT) in 1998 was a major technological breakthrough that dramatically changed the practice of CT. Multiphase imaging can now be performed routinely and with near isotropic resolution (isotropic resolution refers to equal resolution in the x, y, and z axes, so that voxels are cubic in shape), allowing generation of exquisitely detailed three-dimensional volumetric reconstructions. Three-dimensional data sets, narrow collimation, and multiphasic imaging provide improved lesion detection, multiplanar capability, and the ability to perform high-quality CT angiography (Figure 1). With this new technology, which continues to evolve, come significant new challenges, particularly with respect to managing the vast quantity of image data that is generated. This remains an unsolved problem that will require the development of innovative image processing and viewing strategies. The traditional paradigm of reviewing tomographic slices may well be replaced by a primarily volumetric approach. The rapidity with which body cavities can be imaged with MDCT also requires new thinking with respect to the rate of intravenous contrast administration, contrast injection duration, contrast bolus volume, and scan delays. With scanning times as short as 10 seconds, smaller contrast volumes injected at a faster rate may be more appropriate. While narrow collimation may seem intuitively advantageous, it does increase image noise and may reduce the geometric efficiency of the detectors. Finally, the radiation dose is increased with MDCT (although some factors can be adjusted to limit this dose increase), which is compounded by multiphasic imaging. Careful attention must be paid to technical parameters during scanning, with tailoring of the dose and imaging protocol to the clinical question. Use of commercially available dose-modifying software may help. In addition to these technical considerations, a more practical challenge raised by the rapid evolution of MDCT equipment is the choice of scanner. Early MDCT scanners had 4 rows of detectors; subsequent versions offered 8, 12, and 16 rows; and currently 64-row scanners are commercially available. The primary focus of the remainder of this article will be the technical and clinical factors that should be evaluated when choosing between these available options.

Figure 1a. Three-dimensional reformatted CT arteriogram obtained in 1996, using a then state-of-the-art single-detector spiral CT scanner. Mild stenosis (arrow) of the right renal artery is seen. Figure 1b. Surface-rendered view of a CT arteriogram obtained in 2004 using a 16-row MDCT scanner. The information provided by such high-quality CT angiography is competitive with the detail provided by conventional catheter angiography, and the huge technological advance since the era of single-detector CT angiography is self-evident. (Click the image for a larger version.)


The technical improvements that have allowed the development of MDCT include slip-ring technology, high-power x-ray tubes with better cooling, and reconstruction algorithms that can account for nonlinear data acquisition by interpolation. With respect to the number of detector rows, a greater number of rows provides two technical advantages: namely, greater spatial resolution in the z axis and greater temporal resolution due to a greater width of the x-ray beam arriving at the detectors. Faster gantry rotational speeds of modern scanners also add to the increased speed afforded by multiple detector rows with many commercial scanners now capable of a complete gantry rotation in 0.5 seconds or less.

There are some drawbacks to MDCT, for instance, the greater width of the x-ray beam required for MDCT imaging reduces geometric efficiency and contributes to increased radiation dose as a portion of the beam (the penumbra) falls beyond the active detectors and is “wasted.” Such “cone-beam” effects (ie, as the wider beam required to cover more detectors is more cone-shaped than a tightly collimated beam used to radiate a single row of detectors) also complicate the reconstruction algorithms, since the data acquired at the edge of the beam represents a (slightly) oblique projection, whereas the data from the central detectors is planar. Fortunately, modern sophisticated reconstruction algorithms can adjust for these effects. A further drawback to consider is the speed of the table-top. The fastest commercially available MDCT scanners have rotational speeds of up to 0.33 to 0.35 seconds. When combined with high pitch, the table speed can become fast enough to produce a “slingshot” effect on the patient, and becomes a practical technical concern.

Another technical aspect to be considered is the method by which a given vendor has achieved a multidetector design and how many data channels are available. For example, a uniform detector size of 1.25 mm in 16 rows results in an array that is 20 mm in z-axis length, but such an arrangement can be considered a 16-row detector only if there are 16 data channels provided to collect the data. Early 4-row (more correctly, “4-channel”) MDCT scanners had such an array, but had only 4 data channels, so the full potential of the detector array was not realized. Newer 16-detector (really, 16-channel) MDCT scanners typically have a nonuniform detector array, with smaller detectors in the center. For example, one vendor (Vendor A) uses an array of 16 x 0.0625 mm detectors flanked on both sides by 4 x 1.25 mm detectors (for a total of 24 rows, but considered a 16-slice scanner because the data is collected by 16 channels).

The recent commercial release of 64-slice CT scanners introduces additional technical complexity. In the preceding paragraph, it was explained how the number of detector rows can exceed the stated number of “slices” for a MDCT scanner, since the limiting factor is the number of data channels (ie, 16-row array feeding into four channels is conventionally termed a “4 slice” or “4 row” MDCT scanner). While Vendor A achieves 64-slice scanning by having a 64 x 0.0625 mm detector array, Vendor B has taken an innovative approach to 64-slice design where the number of slices is greater than the number of detectors. A floating or wobbling focal spot is used to create two slightly different beam projections on an array of detectors (32 x 0.6 mm central detectors flanked on both sides by 4 x 1.2 mm detectors) feeding 32 channels; the combination of two focal spots with 32 detectors results in a “64 slice” system. Vendor B’s system can achieve rotational speeds of 0.33 seconds, compared to 0.35 seconds for Vendor A. However, the Vendor A system has greater beam thickness (64 x 0.625 = 40 mm versus 32 x 0.6 mm plus 8 x 1.2 mm = 28.8 mm). Finally, these two vendors also use different detector technology, with various claims of superiority. In practice, the choice between such high end machines can be very complex, because the design differences are sufficient to make direct comparison difficult.


Figure 2a. Coronary CT arteriogram obtained with a 64-slice MDCT scanner (64 x 0.625 mm detector thickness, 0.35 gantry rotation time, and 147 mm coverage in 6.5 seconds). CT applications can be divided into those requiring high, intermediate, and low resolution. Coronary CT arteriography is the most technically demanding study that is currently performed using MDCT, requiring high spatial and temporal resolution. Modern 64-slice MDCT systems are capable of producing coronary CT arteriograms that rival catheter angiography. (Click the image for a larger version.)
Figure 2b. Virtual colonoscopy performed on an 8-slice MDCT scanner, showing the right colon in a patient in whom conventional optical colonoscopy had been unsuccessful. Virtual colonoscopy requires intermediate resolution, and can be performed with good results on 8- to 16-slice systems.
Figure 2c. Axial CT image obtained with 5 mm collimation on a 4-row MDCT scanner in a patient with acute diverticulitis. Routine abdominal imaging does not require particularly high spatial or temporal resolution, and most studies can be performed with 4-row technology. (Click the image for a larger version.)

A fundamental consideration when choosing the number of rows or slices for a MDCT scanner is the resolution (both spatial and temporal) required. This is determined by the clinical workload, which can be categorized as studies needing high, intermediate, or low resolution (Figure 2). Undoubtedly, coronary CT arteriography is the most technically demanding study that is currently performed using MDCT. Both high spatial resolution (for visualization of small arteries) and high temporal resolution (visualization during the “inactive” portion of the cardiac cycle) are prerequisites for optimal imaging. A department that performs coronary CT arteriography should there fore choose a scanner that provides the best available combination of spatial and temporal resolutionin practical terms, this means buying a 64-slice system. Studies requiring intermediate resolution include noncoronary CT arteriography and virtual colonoscopy. These studies do require high spatial resolution, but are not as demanding as coronary work in terms of temporal resolution; 8- or 16-slice systems would appear adequate to perform such cases. Finally, low resolution studies would include routine imaging of the chest, abdomen, and pelvis, for example, for cancer staging or acute abdominal pain. Such a workload could be handled quite well by a 4-slice scanner.

The contention that routine imaging of the chest, abdomen, and pelvis does not require high resolution may be controversial, since such studies can be performed with thin collimation and near isotropic resolution. The real question is whether these studies need such thin collimation. There is a reasonable body of evidence, at least with respect to hepatic imaging, that suggests the answer to this question is no. In a study of 31 patients with hepatic metastases (30 from colorectal carcinoma and one from gastrointestinal stromal tumor), preoperative CT showed 88 small hypoattenuating hepatic lesions and 25 of these lesions were metastases. 1 Thinner collimation (2.5 versus 5 mm) improved overall lesion detection, but did not increase detection of these small metastases. If small metastases were isoattenuating, then thinner collimation would not improve lesion detection. Partial volume averaging would also tend to obscure visualization of small metastases that are only slightly different in attenuation to the adjacent hepatic parenchyma, while small cysts are more likely to be seen because of the greater attenuation difference between fluid and hepatic parenchyma. That is, thinner collimation may detect more lesions, but not necessarily more clinically relevant lesions.


Multislice CT represents a major advance in CT technology. Choosing between the spectrum of technology currently available can be simplified by recognizing that the key to selection is the expected workload. Low resolution studies can be adequately tackled by a 4- or 8-row MDCT system, while coronary CTA requires the highest possible resolution, currently represented by 64-slice systems. Given that most practices perform a mixed workload, a strong argument can be made for having a mixture of technologies extending from a 4- or 8-row scanner up to a 64-slice scanner. In such an environment, patients can be assigned to the appropriate scanner depending on the indication. In addition, a multiunit center provides greater efficiency in terms of throughput. It should be noted that irrespective of the choice of scanner or scanners, CT and the associated image storage and postprocessing tools continue to evolve, and capital budgets should reflect the need for product maintenance, upgrades, and replacement in a timely fashion.

Fergus V. Coakley, MD, is associate professor of radiology and chief, Abdominal Imaging Section.

Bonnie N. Joe, MD, PhD, is assistant professor of radiology, Abdominal Imaging Section, Department of Radiology, University of California San Francisco.


  1. Haider MA, Amitai MM, Rappaport DC, et al. Multi-detector row helical CT in preoperative assessment of small (< 1.5 cm) liver metastases: is thinner collimation better? Radiology. 2002;225:137-142.