| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Commentary |
ImPACT, Department of Medical Physics & Bioengineering, St George's Hospital, London SW17 0QT, UK
A new generation of X-ray CT multislice scanners was launched over 2 years ago. Since then their acceptance into clinical practice has been rapid and widespread, and the range of diagnostic applications has expanded. It is anticipated that within a few years over half the CT scanners in the UK will be multislice systems.
Overview of technology
Multislice (or multidetector array) CT scanners, as the name suggests, are capable of acquiring several tomographic slices in a single rotation of the X-ray tube and detector assembly [1]. They share the majority of design characteristics with their single slice counterparts, but their distinguishing feature is the extension of the detector array along the patient's length (z-axis direction). The maximum total imaged length in a single rotation varies between 20 mm and 32 mm on different scanner models. On current multislice systems a maximum of four slices of varying widths can be acquired simultaneously. To achieve this, detector arrays with between 8 and 34 rows are used. For the narrowest slices only the central rows are irradiated. As the X-ray collimation in the z-direction is extended, the signals from adjacent rows are combined to give wider slices.
Multislice scanners can be divided into two broad categories. The fixed or symmetric matrix type has detector elements that are essentially of equal length in the z-direction. Adaptive or asymmetric matrix scanners have detector elements that increase in length with distance along the z-axis from the centre of the array. The advantage of the fixed matrix design is that it allows an easier upgrade path for the acquisition of more than four slices per rotation. The adaptive matrix design is theoretically more dose efficient, having fewer elements and therefore fewer gaps between the elements. However, in practice, the gaps between elements are relatively small and the z-axis geometric efficiency is more dependent on the extent of irradiation outside the active detector length [2].
On single slice scanners the minimum slice width is generally 1 mm, which is determined by the focal spot dimensions and the scanner geometry. On multislice systems, because the X-ray beam is divided between a number of slices, imaged widths as low as 0.5 mm can be obtained. An additional benefit of multislice scanners is the improved utilization efficiency of the X-ray beam in the z-direction; the same X-ray tube loading results in up to four times the volume coverage, allowing increased volumes to be scanned without tube cooling restrictions.
A concurrent development utilized on multislice scanners is the increased rotation speed of the X-ray tube and detector assembly. All multislice manufacturers now market models with a 0.5 s rotation. In some cases the X-ray tube design has been modified to provide greater stability in view of the large centrifugal forces generated at the high rotation speeds [3].
Multislice scanners therefore offer the following improvements over single slice scanners: imaging up to four times the volume in a single rotation; rotation speeds 50–100% greater; minimum slice width halved; and improved X-ray utilization. These advantages, when combined, offer many benefits in a wide range of clinical examinations.
Helical scanning
The majority of examinations on multislice scanners are performed in helical mode and issues regarding pitch are still an area of some debate onthese systems. On single slice scanners thenominal slice width selected determines the X-ray collimation in the z-axis. This defines the imaged slice width in axial scanning, but in helical scanning the imaged width is broadened to some extent when pitches greater than 1 are used. On multislice scanners the parameter "data acquisition width", sometimes termed "detector collimation", must be considered. This is the z-axis extent of detector elements used for the acquisition of each individual slice. In non-helical multislice scanning it is the data acquisition width that generally determines the imaged slice width, although slices can be reconstructed that are multiples of the data acquisition width. In helical multislice scanning the nominal slice width selected is not dependent on the data acquisition width or pitch.
Current multislice systems employ the conventional filtered back-projection methods used on single slice scanners for image reconstruction. In helical multislice scanning there is a difference in the interpolation approach used to create axial data from the helical scan. On most single slice systems, two data samples are always used in the interpolation, resulting in constant image noise with changing pitch. The benefit of reduced dose is obtained at higher pitch values, but at the cost of a wider imaged slice and increase in helical artefacts [4].
On multislice scanners the relationship between pitch and image quality is not as straightforward owing to the variation in z-axis distribution of data samples, created by the interleaving of adjacent multiple helices, at different pitch values [5]. Although there is still a trend towards increased helical artefacts with increasing pitch, certain pitch values result in improved z-sampling distributions. In contrast to single slice scanners, multislice models do not usually use a fixed number of data samples in the helical interpolation, but instead the interpolation is performed over a fixed distance. This approach is known as the z-filter method and a given z-filter distance results in a similar slice width regardless of data acquisition width or pitch [5, 6].
Manufacturers have chosen different approaches to helical scanning to simplify the issue of pitch selection for the user. One approach maintains a constant slice width with pitch by fixing the z-filter distance and on some systems also applying variable weighting functions in the helical interpolation. In contrast to single slice scanners, slice width and noise for a given dose remain constant with pitch using this technique. However, the z-axis sampling density varies and this affects the magnitude of helical artefacts.
Some manufacturers claim that there are no preferred pitches with their approach, whereas others recommend specific pitch values. One manufacturer allows the use of only two fixed pitches, claimed to be optimized with respect to image quality [6]. It should be noted that the view on optimum pitch differs between manufacturers [5–7] and it remains to be shown whether specific pitch values offer significantly improved clinical information. However, as for single slice scanners, improved image quality is achieved for a given imaged slice width by employing a narrow data acquisition width with an increased pitch, rather than vice versa [4].
Manufacturers have not reached a consensus on the definition of pitch, and two different definitions are being used [2]. One is based on the couch travel per rotation relative to the data acquisition width of each slice. This definition has sometimes been termed "detector pitch". The second definition is based on the couch travel per rotation relative to the total active detector width and is sometimes termed "collimator pitch". On a four slice scanner the former definition will result in pitch values four times that of the latter and, when used, highlights the scanner's ability to perform faster examinations. It is therefore the definition that has been adopted by three of thefour multislice manufacturers. However, the second definition is advocated here because, for the same amount of overlap or gap between consecutive spirals, it yields the same values of pitch on multislice and single slice scanners (e.g. collimator pitch=1 relates to the situation where adjacent spirals are contiguous, and collimator pitch<1 relates to overlapping of consecutive spirals). A similar relationship between pitch, dose and level of artefact is therefore maintained.
Image quality
Basic image quality specifications do not differ greatly between multislice and single slice models. Cone beam geometry, resulting from extension of the detector array along the z-axis, can give rise to artefacts when conventional filtered back-projection techniques, as used for fan beam geometry, are employed for image reconstruction [8]. However, on current multislice systems these artefacts have not been shown to be clinically significant.
Improvements in image quality are achieved as a result of the shorter examination times and the resultant decrease in patient motion artefacts. The decreased examination time and the improved X-ray utilization result in the ability to scan routinely with narrower slices. This leads to improved z-axis resolution and a reduction in partial volume artefacts. By selecting the appropriate slice width to match the scan plane resolution, CT images can now approach isotropic resolution, giving rise to high quality three-dimensional (3D) and multiplanar reconstructions. Multislice systems make much use of these modes of display because they can provide increased diagnostic information, and the large number of images created makes viewing of examinations on a slice-by-slice basis time consuming. Reporting of scans from multislice systems is therefore more sensibly performed from workstations.
Helical multislice scanning with narrow data acquisition widths also offers possibilities for combined examinations where reconstructions are performed retrospectively with wide and narrow slices. The wide slices are used for viewing transaxial slices where low noise is a requirement, and high quality 3D or multiplanar reconstructions are obtained from the data acquired with the narrow slices.
Dose
Multislice scanners employ the same detector material, and have the same gantry geometry and X-ray beam filtration as their single slice counterparts. The main difference lies in the z-axis geometric efficiency, where the irradiated slice width is generally wider than the active detector width. This is to avoid calibration problems that can occur when the outermost elements of the active array are exposed to the X-ray beam penumbra.
Varying claims have been made regarding doses from multislice scanners. An early version of one multislice slice model resulted in doses, as measured by the computed tomography dose index (CTDI), that were up to three times higher than a single slice model from the same manufacturer [4]. This high dose resulted from the extent of the X-ray beam beyond the active detector length and reinforced the view that multislice scanners give rise to increased doses. On current multislice models however, when the widest collimations are employed, the CTDI for a given tube current–exposure time product (mAs) is approximately 10% higher than for single slice scanners, but because the z-axis geometric efficiency decreases with total X-ray collimation, for 1 mm slices the values are about 40% higher than for well collimated single slice systems [9, 10]. The view that multislice systems will result in increased doses is also supported by their ability to perform a wider range of applications, the possibility of scanning longer volumes and their capability of performing multiphase contrast studies.
Conversely, it has been claimed that multislice CT will lead to a reduction in dose levels. In support of this view, it is claimed that the shorter examination times reduce the need for repeat examinations caused by patient movement or incorrectly timed contrast studies.
On balance, the extended scope of examinations and reduced z-axis efficiency will probably outweigh the potential for dose reduction and will result in an increase in population dose from CT examinations. However, the greater emphasis on justification and optimization of procedures imposed by the recent EU Directive on Medical Exposures [11] should mean that any such increase is fully warranted.
Conclusion
Multislice scanning offers many clinical benefits. The reduced examination times present advantages, particularly in examinations where voluntary or involuntary patient motion is a problem. These include paediatric, geriatric, trauma, thoracic and cardiac studies. The routine use of narrow slices results in improved z-axis resolution and is of particular benefit in examinations employing 3D reconstructions, such as CT angiography and virtual endoscopy.
We are currently experiencing the early days of multislice CT scanning. Some manufacturers have already announced preliminary dates for the availability of 8-slice and 16-slice models. Other future developments will probably include further extension of the detector array along the z-axis and still faster rotation speeds. The ultimate, long-term aim of all CT manufacturers is likely to be the imaging of the whole trunk in a single rotation [8].
Footnotes
ImPACT is the UK's CT scanner evaluation group, funded by the Medical Devices Agency. 
Received for publication November 2, 2000. Revision received April 10, 2001. Accepted for publication April 30, 2001.
References
This article has been cited by other articles:
 |
|
 |
|
  C. J. Martin RADIATION DOSIMETRY FOR DIAGNOSTIC MEDICAL EXPOSURES Radiat Prot Dosimetry, March 28, 2008; (2008) ncm495v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P C Shrimpton, M C Hillier, M A Lewis, and M Dunn National survey of doses from CT in the UK: 2003 Br. J. Radiol., December 1, 2006; 79(948): 968 - 980. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. L. Storto, A. Di Credico, F. Guido, A. R. Larici, and L. Bonomo Incidental Detection of Pulmonary Emboli on Routine MDCT of the Chest Am. J. Roentgenol., January 1, 2005; 184(1): 264 - 267. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W K Chooi and S K Morcos High resolution volume imaging of airways and lung parenchyma with multislice CT Br. J. Radiol., December 1, 2004; 77(suppl_1): S98 - S105. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S J Yates, L C Pike, and K E Goldstone Effect of multislice scanners on patient dose from routine CT examinations in East Anglia Br. J. Radiol., June 1, 2004; 77(918): 472 - 478. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J Garvey and R. Hanlon Computed tomography in clinical practice BMJ, May 4, 2002; 324(7345): 1077 - 1080. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
