British Journal of Radiology (2004) 77, S2-S9
© 2004 British Institute of Radiology
doi: 10.1259/bjr/28755689
Scanning protocols for multislice CT scanners
E Kulama, MSc
Radiological Sciences Unit (RSU), Hammersmith Hospital, Du Cane Road, London W12 0HS, UK
 |
Abstract
|
|---|
The new multislice CT (MSCT) technology enables examinations to be completed with much greater flexibility in terms of shorter times and improved z-axis resolution. For this to be achieved, a very large amount of data may be produced, with implications on the way the data are handled and reviewed. In addition, there is the issue of radiation dose to the patient and the effect that multislice scanning has on it. In this paper, the basic physics principles of MSCT are reviewed and clarification is given why and where these differ from single slice scanning. Examples of scanning protocols are given with reference to patient dose. These protocols vary depending on the make and model of the scanner and the type of examination.
|
Introduction
|
|---|
Technical developments in X-ray tube power, detectors and computer capacities in the 1990s pushed CT development a step further and led to a renewed interest in it, with multislice CT (MSCT) scanners being introduced by all major manufacturers.
The Elscint-Twin scanner introduced in 1994 was the first system with a dual detector array, capable of acquiring two images simultaneously. 1998 saw the introduction of 4-slice systems with 0.5 s scan times. This meant that the examination times could be reduced up to a factor of 8 compared with the single slice, 1 s systems [1].
One obvious advantage of shorter scan times is the reduction of motion artefacts caused by patient movements, leading to an improvement in image quality.
All commercial multislice scanners today are 3rd generation scanners – scanners with rotating X-ray tubes, generators and detector arrays.
|
Detectors
|
|---|
Once the ionising radiation exits the body, it is recorded by the detector system, which transforms it to an electrical signal. The signal is then amplified, converted from analogue to digital form and can be reconstructed to form an image in the same way as in single slice CT, by filtered backprojection.
The detectors used in MSCT scanner systems are ceramic scintillation detectors (such as gadolinium oxysulfate).
The designs of the multirow detector arrays that manufacturers are using for their 4-slice systems are shown in Figure 1
.
GE used a 16-row detector array, with each detector being 1.25 mm wide at the isocentre. This gives the detector a total length of 20 mm and the combinations of slice widths that can be acquired simultaneously are: 2 x 0.63 mm, 4 x 1.25 mm, 4 x 2.5 mm, 4 x 3.75 mm and 4 x 5 mm. Siemens and Philips use the same detector array, which consists of eight rows of detectors of varying widths with a total length at the isocentre of 20 mm. This design is known as an adaptive array and the combinations of slice widths possible are: 2 x 0.5 mm, 4 x 1 mm, 4 x 2.5 mm and 4 x 5 mm. Toshiba's detector array is a 34 row detector array (with a total length of 32 mm at the isocentre). There are four 0.5 mm detectors in the centre and fifteen 1 mm detectors on either side. The combinations of slice widths in this detector are: 4 x 0.5 mm, 4 x 1 mm, 4 x 2 mm, 4 x 3 mm, 4 x 5 mm and 4 x 8 mm.
With multislice arrays, the slice width is determined by the size of the detector element used. When the slice width is greater than the detector element, the outputs from a number of elements along the z-axis are summed.
Since the introduction of 4-slice scanners, manufacturers have developed scanners capable of acquiring more and more simultaneous slices. Currently there are commercially available scanners than can acquire from 2 up to 16 slices simultaneously and there is work in progress to increase these numbers.
|
Image reconstruction
|
|---|
Axial image reconstruction in MSCT follows the same general principles as in single slice helical CT; in helical MSCT however, the user can select an effective slice width, equal or larger – never smaller – than the acquired slice width. The smaller the acquired slice width and reconstruction increment, the finer the sampling along the z-axis and therefore the better the spatial resolution along that direction.
The ability to image very thin slices (<1 mm) improves the z-axis resolution, giving almost isotropic resolution. This enables reconstructions not only in the axial plane but also in the sagittal and coronal orientation without the need to employ special positioning techniques. Reconstructions in these orientations are built up by putting together the same image row or line from a series of successive axial images. Images can be selected in arbitrary orientation and the only efficient way to view and diagnose the large image data sets is by interactive evaluation on soft copy. In addition, three-dimensional (3D) displays representing the scan volume can be reconstructed with a high degree of accuracy. These can provide the anatomical reference points in one image and can also be valuable for the referring physician or the surgeon.
However, when scanning with thin, submillimeter slices the mAs is usually increased to keep the noise in the image low. This will also increase the patient dose.
As the number of slices increases from 1 to 4 to 16, the geometry of the slice changes from being a planar disc towards being more like a cone, especially for the outer slices in the detector bank. When using standard reconstruction algorithms, this leads to streaking artefacts. Specialized reconstruction algorithms are used for 16-slice scanners to help overcome this.
|
Helical pitch
|
|---|
Pitch is a parameter that was introduced with helical scanning and is the ratio of the table feed (in mm) per rotation to the total slice collimation. It is a dimensionless quantity and has an important effect on image quality and dose. On MSCT scanners the pitch can vary between 0.5 and 2. If the pitch is less than 1, overlapping scanning occurs and therefore the dose is increased, although unlike in single slice scanning, the image quality will improve. With pitches between 1 and 2 the scan volume is covered faster and the dose is lower than sequential scanning, but with an increase in image noise and interpolation artefacts. Some scanners will automatically compensate for the increase in pitch by increasing the mA to keep the image noise constant. The choice of pitch will depend on the detector array, rotation times and available algorithms for image reconstruction.
|
Image quality
|
|---|
As with every diagnostic modality, the quality of the image in CT is of primary importance. Images suitable for their intended diagnostic purpose are required, therefore low noise, high contrast resolution, sharpness of the image and the absence of artefacts would be the ideal. However, this is not practical, since patient dose and scan time have to be taken into consideration. In addition, reconstruction parameters – convolution kernels, reconstruction increments, effective slice width, z-interpolation algorithms, pitch – all affect the image quality and all vary depending on the manufacturer and the scanner model.
|
Dose to patient
|
|---|
CT scanning is capable of providing high quality diagnostic information but it is also usually described as being a high dose procedure. With the introduction of MSCT, the performance capability of the scanners has increased dramatically and a wider range of examinations can now be performed. Shorter scan times, multiphase protocols and much thinner slices are all now possible. A survey by the National Radiological Protection Board (NRPB) in 1999 showed that 4% of all diagnostic examinations are CT examinations and they contribute approximately 40% of the patient dose [2]. There is every indication that these numbers are increasing.
Absorbed dose is the energy absorbed per unit mass and is measured in Gray (Gy) and its subunit the milliGray (mGy).
Absorbed dose in CT is characterized using the Computed Tomography Dose Index (CTDI). This is the mean dose within a scanned medium. CTDI is a general definition and can be measured under a variety of circumstances. For example, it can be measured in air or in a Perspex phantom of 16 cm (representing head) or 32 cm (representing body) diameter. All of these values are the CTDI but they serve different purposes. Within Perspex phantoms, measurements are commonly made at both the centre and the periphery (1 cm from the surface). The weighted sum of central and peripheral CTDI values is known as the weighted CTDI or CTDIw and represents the mean dose in the x–y plane. CTDIw is a useful concept for contiguous axial scanning, but in helical scanning the pitch must be taken into account to give the mean dose within a scanned volume, CTDIvol. Another parameter used is the dose–length product (DLP). DLP is related to the total radiation exposure, as it is the CDTIvol multiplied by the length of the volume scanned. DLP is measured in (mGy cm) and gives an approximation of radiation risk but it cannot give a direct assessment of the patient dose, since it does not take specific organs into account [3].
Since May 2000, assessment of patient dose in CT is required by the IR(ME)R2000 legislation (http://www.doh.gov.uk/irmer.htm) [4]. An accepted way to monitor patient doses is to set diagnostic reference levels (DRLs) for each examination protocol, which, when exceeded, require justification. DRLs in CT are described in terms of CTDIvol (mGy) and DLP (mGy cm).
Established DRLs for certain examinations can be found in the "European guidelines on quality criteria for computed tomography" (EUR 16262, May 1999; http://www.drs.dk/guidelines/ct/quality/htmlindex.htm). This data set is based on a survey from the late 1980s and therefore the NRPB with the help of ImPACT and the CT Users Group are currently conducting a patient dose survey (http://www.impactscan.org/dosesurveysummary.htm) to establish newer national DRLs for the UK. In addition, DRLs must be set at a local level within every Trust that uses CT scanning.
The CTDIvol and/or the DLP are now displayed on the scanner's operating console prior to scan acquisition and can be used to optimize the dose delivered by a prescribed CT protocol. These values may also be noted on the patient's records for comparison with UK or local DRLs.
The basic dosimetry characteristics of any scanner are related to the X-ray tube, the beam filtration and the scanner geometry and are very similar for multislice and single slice scanners. There is, however, a technical difference between single slice and multislice scanners in respect to dose. To ensure that all detector rows receive a signal of the same intensity, a similar primary beam intensity must reach all detectors. As a result, the fan beam must be widened along the z-axis because the penumbra is less intense. This introduces a reduction to the z-axis geometric efficiency of the system (i.e. the percentage of useful beam width over total beam width that reaches the detectors). The reduced geometric efficiency is more significant for narrower collimations since, as shown in Figure 2, the penumbra region is more significant for the 4-slice rather than the 8-slice acquisitions, resulting in reduced geometric efficiency.
The patient dose depends on the scanning parameters and the body region examined, but also on patient size. There is a linear relationship between tube current (mA), scanning time (s) and dose. In addition, the same relationship exists between scanning volume and DLP, although in this case the organ and effective dose can vary dramatically depending on the body region being scanned. By increasing the pitch for the same scanning volume, the dose is decreased, although many systems now compensate for this by increasing the mA.
In all helical examinations, in order to gather the required data for reconstruction, the scanner will irradiate a larger volume than is actually being imaged. For single slice scanners this can be either a half or full rotation at each end of the scan. For multislice scanners this can be greater than one rotation and, owing to the wider beam, it becomes more significant, particularly in short scan volumes.
Conscious efforts to reduce patient doses must be implemented by (i) the user, who must select the appropriate protocol and scanning parameters for all CT examinations and (ii) the manufacturer, who must develop dose efficient systems and default protocols as well as good user training.
|
Examination protocols
|
|---|
CT is a dynamically developing diagnostic technology. MSCT scanners are capable of acquiring 2, 4, 6, 8, 10 or even 16 slices per rotation, with tube rotation times as fast as 0.5 s. This enables examinations to be performed within a breath-hold, or the whole body to be imaged in a very short period of time. By being able to image the whole body quickly or by scanning additional phases in multiphase studies, the dose to the patient may be increased.
For any examination to be justifiable, the risks of radiation exposure that patients face must be outweighted by the clinical benefits. These risks, although small, are still a concern. It is important to recognize that the potential biological effects from radiation depend not only on radiation dose, but also on the biological sensitivity of the tissue or organ irradiated.
We must also remember that children require only a fraction of the exposure parameters administered to adults and therefore paediatric protocols must be adjusted accordingly. Becoming acquainted with the dose issues related to MSCT is key to reducing the risks of unnecessary exposure.
The dose to the patient and the scan protocols can and should be optimized. Some of the best strategies available for reducing radiation dose are: (i) technique chart utilization to allow for mAs reduction in relation to the patient's size and weight; and (ii) implementation of automatic exposure control systems by the manufacturers.
|
Protocol optimization
|
|---|
Image quality criteria for all examination protocols must be set. In an ideal world where dose to the patient would not be an issue, the maximum image quality would be the only goal. In reality, however, a compromise must be reached between dose and image quality.
When first faced with a MSCT scanner, the number of available options can be overwhelming and setting up a protocol from scratch is not an easy task. This is where the pre-loaded manufacturers' protocols can help. They have already been tested in a clinical setting and take into account the specific scanner's image quality and dose characteristics.
Different multislice scanners, from Siemens and GE (Table 1), that employ different standard protocols are used in this article as an example. Table 2 refers to head protocols, Tables 3 and 4 refer to high resolution and routine chest protocols and Table 5 refers to routine abdomen protocols.
Variations in the protocols reflect each scanner's capabilities to select a particular range of kV, mA and rotation time values and to apply different reconstruction algorithms.
Note that when comparing different CTDIvol values for each protocol, it is important to realize that the image quality levels may also vary.
Examinations on each make and model of MSCT scanner can be set up by simply selecting one of the pre-loaded protocols. For each protocol there are a number of parameters to consider, as seen in Tables 2–5
.
- Axial or helical scanning: the main advantage of MSCT scanners is their ability to scan large volumes with short rotation times, particularly with helical examinations. For this reason, in the majority of cases the helical mode will be selected. There are, however, cases such as high resolution lung studies or posterior fossa trauma or stroke scans, where axial scanning might be preferred. In the former case, only a few slices through the lungs are enough to indicate interstitial changes and, by not irradiating the whole volume, the dose to the patient is significantly reduced. In the latter case, axial scanning is used to reduce artefacts. However, MSCT enables scanning with acquired data width of 1 mm or less and therefore the resolution along the z-axis is much improved and the artefacts are reduced.
- kV: selection of tube voltage will depend on the scanner make and model. The kV range is 80–140 kV, with lower kV producing higher image contrast and higher kV enabling lower doses to be used.
- Rotation time (s): this can vary between 0.5 s and 4 s depending on the scanner.
- mA or mAs: the range of tube current (mA) values, or its product when multiplied with the tube rotation time (mAs), will again depend on the scanner make and model. The mA range is usually between 10 mA and 500 mA. The higher the mAs the less noisy the image but also the higher the dose to the patient.
- Slice collimation: this is the total X-ray beam thickness and is equal to the number of selected detector channels multiplied by the detector channel width. The range of slice collimations will depend on the detector array used in each scanner and varies between 1 mm and 32 mm.
- Slice width: this is the width of the reconstructed image. The acquired data set can be reconstructed at a range of thicknesses equal to or larger than the original collimation (Table 6). Remember that the wider the reconstructed slice, the lower the noise, but also the lower the z-axis resolution of the images. By creating multiple retrospective reconstructions with different slice widths, both high z-axis resolution and low noise image sets can be provided without additional exposure to radiation.
- Feed/rotation: the table movement (in mm) per rotation.
- Pitch factor: the table movement per rotation divided by the slice collimation, e.g. the table moves 27.5 mm per rotation, the slice collimation is 16 x 1.25=20 mm, therefore the pitch factor is 1.375. In many systems the pitch factor is not selected directly; once the slice collimation and feed per rotation are selected the pitch factor can be calculated.
- Image interval/increment: this is the image reconstruction interval; it will affect the number of slices but not the patient dose.
- Kernel/reconstruction algorithm: these will depend on the make and model of the scanner and also on the body part selected. Different reconstruction filters can be used for varying degrees of smoothing or edge-enhancing the images. Reconstruction parameters can be defined either before and/or after data acquisition.
- CTDIvol: the values displayed in the tables have been calculated using the ImPACT CT Dosimetry Spreadsheet (http://www.impactscan.org/ctdosimetry.htm).
View this table:
[in this window]
[in a new window]
|
Table 6. Possible reconstructed slice thicknesses in axial and helical scanning for the Siemens Sensation 16 scanner
|
|
Another essential measure is to adapt the scanning parameters to the size/diameter of the patient. This is particularly true for paediatric examinations, where the mAs values used can be significantly lower than the adult ones. Paediatric protocols must be set up to reflect paediatric sizes and anatomy. Tables 7 and 8 give examples of paediatric head and chest protocols on the Siemens Sensation 16 scanner.
Siemens as well as Philips and Toshiba modify the paediatric scan protocols for head and neck based on age and the protocols for the other regions according to body weight.
GE scanners have a similar set of paediatric protocols based on children's age and body weight. The protocols are colour co-ordinated with the American "rainbow system", where the acquisition parameters are set for each colour (pink, red, purple, yellow, white, blue, orange, green, black) that represents specific age/body weight ranges.
|
Automatic exposure control (AEC) and mA modulation
|
|---|
All manufacturers have taken on board the need to control the balance between patient dose and image quality. One way to achieve this is with modulation of the tube current related to the anatomy of the volume scanned.
Cross-sections of the human anatomy differ in size from one patient to another and from one body region to another. They are also in the majority elliptical. Therefore, attenuation of the X-ray beam will be different for different body sizes/regions and in the anteroposterior (AP) and lateral directions. As a consequence, when constant exposure settings are used, image quality varies from point to point and within a series. By using automatic control for mA we can compensate for this.
Different manufacturers approach solutions in their own way. GE Medical systems, as part of the "auto mA" software, use attenuation data from the scout view to adjust the tube current in each rotation of an axial, helical or cine scan. This way the noise is constant on all of the images, while the overall dose to the patient is optimized for that image quality. For example, in a chest/abdomen examination, the anatomy changes dramatically from the abdomen through the lungs and the shoulders, but the ability to vary the mA from one body region to the next enables maintenance of consistent image quality at a lower dose (Figure 3). The system requires selection of a new parameter called Noise Index (NI), which relates to the acceptable noise level, i.e. the amount of noise the user can tolerate and still deem the image diagnostic. Depending on the selected protocol (kV, slice thickness, pitch), the appropriate mA per rotation for the desired NI is determined by the scanner. A table showing the mA values per rotation is available for viewing prior to scanning. To ensure that there is a dose reduction in a protocol when using "auto mA", a maximum mA value equal to the fixed mA value for that protocol can be set. Default NI values are incorporated in each protocol but can be overwritten by the user.
Philips use "Dose Right", where "Automatic Current Selection" varies the current for each patient depending on their size as seen on the scout view. This is set by means of a "reference image', whose noise value the system attempts to duplicate.
With their "CARE dose" application package, Siemens continuously modulate the tube current during the rotation for both axial and helical scanning. Based on the feedback signal from the first detector row used in the scan, the current is generally reduced in the AP direction compared with the more attenuating lateral direction. It is therefore possible to reduce the overall dose to the patient without significant reduction in image quality. "CARE dose" is pre-selected by default in all standard protocols but it can be switched on/off and it does not require any changes in the scan parameters. The mean value of the mAs applied per rotation will be lower than the set mAs and it is recorded on each image (Figure 4).
Since the mA is adapted in real-time (based on the patient's attenuation profile), the effective mA values used per scan can only be viewed after the acquisition. Published data show dose reduction up to 22% in the thorax, 15% in the abdomen and 25% in the pelvis when using "CARE dose" [5, 6].
Finally, Toshiba use the "Real EC" to vary the current per rotation. Again, the mA varies along the z-axis of the patient and the acceptable amount of noise in the image can be predefined.
The advantages of all these methods are clear: a more constant signal reaches the detectors and as a result the image quality is also more constant, especially through the shoulders and the pelvis. Artefacts can be reduced while the dose is optimized. In addition, the tube heat capacity can be conserved.
|
Discussion and conclusions
|
|---|
Multislice scanners follow broadly the same design as single slice systems with respect to patient dose. However, attention must be paid to the reduced dose efficiency shown in 8-, 10- and 16-slice scanners, especially at narrow beam geometry.
Owing to the flexibilities of modern systems, examination times can be reduced and an examination can be performed within a breath-hold. Since the systems are so fast, the temptation can be to extend the scanning volume or to add a new phase to an examination. In addition, when performing scans with very thin slice widths, the mA per rotation can be increased to keep the noise to the image in an acceptable range. These alterations from a standard protocol will lead to an increased patient dose.
Unfortunately, owing to the variety of scanner makes and models, it is not possible to recommend a protocol that will work equally well on all. In addition, patient morphology and clinicians' ideas for diagnostic image quality vary enormously. The information presented here is for guidance only; every healthcare practitioner must use their own training and expertise in adapting these basic principles and recommendations for each protocol and patient when necessary.
Increased focus on paediatric doses from CT has led manufacturers to pay special attention in defining paediatric protocols. Doses to children are significantly increased if the scan parameters are not adapted sufficiently.
In an attempt to optimize patient dose and image quality, manufacturers have introduced AEC and mA modulation software on the scanners. Early results are very promising and the use of those software packages is strongly recommended. An easy way to check on the patient dose for each protocol is to compare the displayed CTDI and DLP values for old and new protocols.
Finally, to take full advantage of the isotropic resolution in multislice spiral CT, scanning of volumes with thin slice collimation and overlapping reconstruction is employed. This leads to a large amount of images being produced by scanners that can only be efficiently and conveniently be evaluated by soft copy reporting.
|
Acknowledgments
|
|---|
The author would like to acknowledge the invaluable help of Nicholas Keat from ImPACT, Renato Leite and Susie Guthrie from Siemens Medical, and Sandie Jewell and Jane Hickey from GE Medical Systems.
Received for publication August 12, 2003.
Revision received February 11, 2004.
Accepted for publication February 16, 2004.
|
References
|
|---|
- Kalender WA. Computed tomography: fundamentals, system technology, image quality, applications. Germany: Publicis MCD Verlag, 2000.
- Shrimpton PC, Edyvean S. CT scanner dosimetry. Br J Radiol 1998;71:1–3.[Medline]
- International Commission on Radiological Protection. Managing patient dose in computed tomography, ICRP Publication 87. Ann ICRP 2001;30(4).
- The Ionising Radiation (Medical Exposure) Regulations 2000 (SI 2000 No. 1059). London, UK: HMSO, 2000.
- Hamberg LM, Rhea JT, Hunter GJ, Thrall JH. Multi-detector row CT – radiation dose characteristics. Radiology 2003;226:762–72.[Abstract/Free Full Text]
- Greess H, Wolf H, Baum U, Lell M, Pirkl M, Kalender W, et al. Dose reduction in computed tomography by attenuation-based on-line modulation of tube current: evaluation of six anatomical regions. Eur Radiol 2000;10:391–4.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
A J Van Der Molen, W J H Veldkamp, and J Geleijns
16-slice CT: achievable effective doses of common protocols in comparison with recent CT dose surveys
Br. J. Radiol.,
April 1, 2007;
80(952):
248 - 255.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
