BJR
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Crawley, M T
Right arrow Articles by Wainwright, A
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Crawley, M T
Right arrow Articles by Wainwright, A
British Journal of Radiology 74 (2001),607-614 © 2001 The British Institute of Radiology

Full paper

A practical approach to the first iteration in the optimization of radiation dose and image quality in CT: estimates of the collective dose savings achieved

M T Crawley, MSc, MIPEM, A Booth, MRCP, FRCR and A Wainwright, DCR

Radiology Department, Stoke Mandeville Hospital, Mandeville Road, Aylesbury, Buckinghamshire HP21 8AL, UK


   Abstract
Top
 Abstract
Introduction
 Method
 Results
 Discussion
 References
 
CT is a diagnostic imaging modality giving higher patient dose in comparison with other radiological procedures. It contributed an estimated 20% to the collective effective dose to the UK population from medical X-rays in 1990, rising to an estimated 40% in 1999. Tremendous national effort has been expended in reducing patient doses from other radiological procedures with considerable success, but much of the collective dose reduction achieved has been offset by a corresponding increase in the collective dose from CT. Over a period of about 2 years, following the installation of a helical CT scanner, CT scan parameters in this hospital have been adjusted with the aim of working towards optimization of image quality and patient dose. A 33% reduction in annual collective effective dose has been achieved, from about 16.5 manSv to 11 manSv. However, despite this dose reduction, the annual collective effective dose to our sub-population is 2.8 times the value 9 years ago. The increase is almost entirely the result of an increased application of CT; 6149 examinations per annum in 1999 compared with 2210 in 1991. The crucial importance of reducing patient doses from this modality is discussed. Indicative effective doses and image noise values are presented for examination protocols approaching optimization.


   Introduction
Top
 Abstract
 Introduction
Method
 Results
 Discussion
 References
 
Nationally, the collective effective dose to the UK population from CT increased during the 1990s and could still be increasing, as faster scanners with more sophisticated software allow higher patient throughput and examination of a wider range of pathologies. The contribution of CT to the annual population dose from medical radiation exposures was estimated at 20% in 1990, rising to 40% in 1999 [1, 2]. Physicists, radiologists and radiographers are well aware of the potential for elevated doses to patients from CT examinations and, because of its large contribution to the collective dose, CT presents as an obvious target for dose reduction strategies.

Following the implementation of dose reduction strategies in the late 1980s, the annual collective effective dose from CT at this hospital was previously reported as 3.9 manSv (2210 examinations) in 1991 for a conventional axial CT scanner (GE 9000 HP; General Electric Medical Systems, Milwaukee, WI) [3]. In the present study we have calculated the annual collective effective dose following installation of a helical CT scanner (Siemens Somatom Plus 4; Siemens Aktiengesellschaft, Erlangen, Germany), and again approximately 2 years later, after adjustment of clinical scan protocols, in the first iteration of our work towards optimization of patient dose and image quality.


   Method
Top
 Abstract
 Introduction
 Method
Results
 Discussion
 References
 
CT collective effective dose can be reduced in several ways, the most useful strategies being justification of each individual examination by a consultant radiologist, reduction of the scanned volume, optimum selection of technique factors (kV, mA, rotation time, slice width and pitch (for helical scans) or couch increment (axial scans)). All these strategies were employed in the process of optimizing patient dose and image quality for CT scan protocols at this hospital following the installation of a helical CT scanner. The original scan protocols used were based on those recommended by the manufacturers as a starting point for clinical work, but these were not optimized as the optimization process requires local clinical input.

For each type of scan, agreement was achieved between all reporting radiologists on the scanned volume, with a check being made to ensure that European quality criteria for diagnostic images were met in this respect [4]. This process was led by one radiologist with a special interest in CT. In an iterative procedure, tube current (mA) and/or rotation time were then gradually reduced, and slice width and/or pitch were gradually adjusted, under the supervision of a consultant radiologist. It is essential that such adjustments are made in an incremental fashion in order that unacceptable deterioration in image quality is detected and avoided. Throughout this gradual process of protocol adjustment, CT images were assessed against the image criteria given in the European working document to ensure that diagnosticity was not compromised. In general, the lowest mAs values have been achieved in areas where there is high inherent contrast, e.g. chest, spine and imaging for renal calculi. In almost all body scan protocols it has been possible to adopt a pitch of at least 1.5 and a tube potential of 120 kV. For head scan protocols (all at 140 kV), where higher spatial resolution is required, sequence scanning or a pitch of 1.0 is normally used, together with smaller slice widths. The scanner can also be operated at 80 kV, although to date this tube potential setting has not been used clinically.

Thus, standardized low dose protocols were derived for a range of scanned volumes and pathologies. Table 1Go shows four of our current protocols for some common examinations, namely standard brain, chest, abdomen and pelvis, for this particular model of CT scanner, although in fact there are now over 30 pathology-specific protocols in clinical use. Each pathology-specific protocol defines the pre- and post-contrast scan volumes, slice width, tube potential, tube current and scan time. It has been proposed elsewhere that scan protocols have limited usefulness in determining individual patient doses, as clinical practice may vary to suit individual patients in terms of size and clinical indication [5]. However, because we have developed an extended range of protocols tailored to specific clinical indications, deviations from the protocols only occur in relation to patient size, i.e. tube current is increased for an obese patient, or decreased for a paediatric patient. Our pathology-specific protocols, which relate to all average adult patients, are therefore useful in determining representative patient doses that enable comparison of different scan protocols. Our current scan protocols are now felt to be approaching the optimum, and any further attempts at dose reduction will necessitate a more in-depth analysis of image quality.


View this table:
[in this window]
[in a new window]
  		Table 1. Scan parameters for some common clinical protocols

 
Image noise levels were assessed by determining the statistical variation in the CT number of Perspex, using a 16 cm diameter head phantom and a 32 cm diameter body phantom. The phantoms were scanned using a representative selection of the low dose and original scan protocols. The method used to determine the noise values has been described elsewhere [6].

Effective doses were calculated using the commercially available software CTDOSE for both the original and current protocols [7]. Pre- and post-contrast scan volumes were defined for all protocols, therefore effective doses for both components of a given examination were calculated and then combined to give the total effective dose per examination. CTDOSE uses normalized organ dose data sets, together with measured values of free-in-air axial dose for particular models of scanner, and details of the clinical technique for each type of examination [8]. These organ dose data sets were derived using Monte Carlo calculations in an anthropomorphic phantom representing a 70 kg hermaphrodite adult and currently relate to a previous generation of axial CT scanners, but details of appropriate matches for newer scanners to the original data sets have been produced by ImPACT (a Medical Devices Agency evaluation group) [9]. For the Somatom Plus 4 CT scanner, data set 1 was used for head scans and data set 16 for body scans at both 120 kVp and 140 kVp.

The CTDOSE software requires the following input parameters: scanned volume (in terms of baseline in the phantom and number of slices), slice width, couch increment (axials), effective mAs and CT dose index per mAs (CTDI). CTDI is a measure of the total dose from a single slice irradiation. The European working document gives the following formula for CTDI [4]:Go


where Z1 and Z2 are the limits of integration, D(z) is the single slice dose profile and T is the nominal slice thickness in centimeters. Integration distances and the phantom in which the measurement is made are identified in suffix notation, e.g. CTDI{Delta}Z,phantom, where the suffix {Delta}Z is the integration distance, and "phantom" denotes measurement in a specified phantom or air.

The CTDI10cm,air values used in the dose calculations are shown in Table 2Go and were obtained from measurements using a pencil ion chamber (active length 10 cm) and dosemeter (type 35050A; Keithley Instruments Inc., Cleveland, OH). The measurements were made free-in-air, with the chamber on the scanner central axis. Because the organ dose data are expressed in terms of absorbed dose to ICRU muscle, these measured values of dose-in-air were converted to dose-to-ICRU muscle using appropriate conversion factors before input into the program [10].


View this table:
[in this window]
[in a new window]
  		Table 2. CT dose index (CTDI) values (mGy mAs-1) for the Siemens Somatom Plus 4 CT scanner

 
As the software relates only to axial slices, it was necessary to find a way of simulating a helical CT scan for protocols using helical scanning. To simulate a fairly uniform irradiation of the scanned volume within the phantom, the effective mAs was divided by the pitch. The value of the couch increment was then taken as being equal to the slice width to emulate contiguous slices, and care was taken to choose a slice number value that ensured coverage of the scanned volume.

Some inaccuracies were introduced for scans with gantry tilt, namely all scans that use a coronal projection (sinuses, facial bones, orbits and petrous bone), as CTDOSE does not cater for this projection. Effective dose for corresponding axial scans was used as an approximation in these instances, with inevitable loss of accuracy. Gantry tilt is also used on a small percentage of lumbar spine scans, where the clinical aim is identification of a pars defect, therefore some loss of accuracy is to be expected here also. It has been shown elsewhere that the approximation of an angled projection to an axial slice can result in errors of up to 20% [11].

The quantity dose–length product (DLP) was then derived for all scan protocols using the methods described in the European working document, for comparison against the four proposed diagnostic reference levels relating to head, chest, abdomen and pelvis [4]. The quantity DLP uses a weighted CTDI, (CTDIw (mGy)). CTDIw is an approximation to the average dose over a single slice and is derived from a combination of measurements at the surface and centre of a defined set of Perspex phantoms, according to the equation:Go


where CTDI10cm,c is the CTDI measured at the centre of the phantom and CTDI10cm,p is the CTDI measured at the periphery of the phantom.

The European working document gives the following formulae for DLP:Go


Go


where i represents each scan sequence forming part of an examination, nCTDIw is the normalized weighted CTDI in the phantom (mGy mAs-1), N is the number of slices, T is the slice thickness, C is the radiographic exposure in mAs, A is the tube current (mA) and t is the total acquisition time for the sequence. The nCTDIw values used for the scanner in this study are given in Table 2Go.

Data on patient attendances for a 12-month period from April 1999 to April 2000 were obtained from the Radiology Information System (RIS), in which CT examinations are allocated to Korner categories [12]. The number of examinations that included a contrast run could be readily determined from the data, as these were logged into separate categories. The Korner data were then used together with the data on effective doses to determine annual collective effective doses for each Korner category. Of a total of 6149 examinations, 100 limb examinations were excluded from the analysis as limbs are not included within the software phantom used by CTDOSE. The effective dose from this category of scan will be comparatively small (<0.1 mSv) and the contribution to collective effective dose is likely to be less than 0.01 manSv. Table 3Go shows how the remaining 6049 examinations in 1999–2000 were allocated to the various Korner categories. Again, some loss of accuracy accompanies this process, as about half the Korner categories contain several different scan protocols. For example, the Korner category "spine" includes patients for cervical, thoracic and lumbar spine, each of which has a different effective dose owing to the different organs irradiated. Furthermore, the exact protocol used depends on the pathology. As the majority of these examinations were T or L spine, the average effective dose for these two examinations was used in determining the collective effective dose in this category. Similarly, other Korner categories, e.g. chest and abdomen, contain scans carried out using a variety of protocols according to pathology. For such Korner categories it was possible to derive approximate effective doses per category by taking a simple average of the effective doses for all the scan protocols within that category.


View this table:
[in this window]
[in a new window]
  		Table 3. Allocation of examinations to Korner categories: examination numbers, effective dose (ED) and collective ED before and after dose reduction

 
Different Korner categories applied in 1991; the number of examinations in the different examination categories for 1991 and 1999 are shown in Table 4Go, together with the percentage increases for head, spine, chest, abdomen and pelvis examinations.


View this table:
[in this window]
[in a new window]
  		Table 4. Examination categories and number of examinations in 1991 and 1999

 

   Results
Top
 Abstract
 Introduction
 Method
 Results
Discussion
 References
 
Effective doses for each scan protocol are shown in Table 3Go, together with the collective effective doses in each Korner category, before and after this first iteration in the dose reduction process. Figure 1Go also shows collective effective doses before and after protocol adjustment. Before protocol adjustment, the total annual collective effective dose to the subpopulation scanned at this hospital was about 16.5 manSv. After protocol adjustment this has been reduced to 11 manSv. Collective dose savings are particularly notable for scans of the chest and pelvis.



View larger version (22K):
[in this window]
[in a new window]
  		Figure 1. Collective effective dose before and after dose reduction.

 
The number of patients referred for CT has increased by a factor of 2.8 since 1991 (2210 patients compared with 6149 patients in the period April 1999 to April 2000). The average effective dose per examination, calculated by dividing collective effective dose by the total number of examinations, has been reduced from an initial value of 2.7 mSv per examination to 1.8 mSv per examination, which is close to the 1991 average of 1.7 mSv. Table 5Go shows a comparison of effective doses for scans in 1991 and 1999.


View this table:
[in this window]
[in a new window]
  		Table 5. Comparison of effective doses (mSv) for standard scans in 1991 and 1999

 
DLP values for all protocols, except brain with contrast, were well below the proposed diagnostic reference levels. Table 6Go shows the protocol with the highest DLP for each of the main protocol groups, together with the proposed reference level for that group.


View this table:
[in this window]
[in a new window]
  		Table 6. Comparison of highest dose–length products (DLPs) against proposed reference levels

 
Apart from the high brain dose, other instances where European Commission (EC) quality criteria were not met were in the occasional use of a 1 mm slice width for pre-operative petrous scans on small children for cochlear implants, and the increase of the scanned volume in cases of vertebral trauma to include one undamaged vertebra above and below the fracture site, as these are used as fixation points during subsequent surgery.

Indicative image noise levels, obtained using a selection of the original and current protocols to scan Perspex phantoms, are shown in Table 7Go.


View this table:
[in this window]
[in a new window]
  		Table 7. Flat field noise levels (%) for current standard clinical protocols in Table 1Go (noise levels for scans using the original protocols in parentheses)

 

   Discussion
Top
 Abstract
 Introduction
 Method
 Results
 Discussion
References
 
Our previous study showed that the annual collective effective dose on an older generation pulsed axial CT scanner at this hospital was 3.9 manSv per year, based on 2210 examinations in 1991 [3]. Following initial dose reduction in working towards opimization, the current value, based on 6149 examinations in the period April 1999 to April 2000, is 11 manSv. This is considerably lower than the National Radiological Protection Board's (NRPB) 1991 estimate of 20 manSv per scanner per year on average, although at the start of the optimization process we were closer to this estimate [10]. The effective dose per examination, averaged across all scan protocols, has returned to about the same level as 1991 (1.8 mSv for 1999 compared with 1.7 mSv for 1991), therefore the 2.8-fold increase in collective effective dose since 1991 is almost entirely the result of increased application of this modality, for example chest referrals have increased 9-fold. Also, the use of CT pulmonary angiography for pulmonary embolism and high resolution chest CT in particular are new techniques following the introduction of the helical scanner. Similarly, CT pneumocolon has been introduced and oral contrast CT bowel has replaced enemas for patients who are over 80 years of age. The oncological workload has increased significantly for pre-operative assessment, staging and follow-up as per national guidelines. If MRI had not been available throughout this period to offset the CT workload, it is likely that the increase in CT referrals would have been significantly greater, although until recently body MRI has not been used routinely. Therefore, despite MRI availability, and the concerted effort by staff to reduce doses, we estimate that CT now contributes 45% to the collective dose from all our diagnostic radiology (100 000 patients per year). This supports the NRPB estimate that escalating CT doses now contribute 40% to the UK collective effective dose from all medical exposures.

The DLP of 1253 mGy cm for a brain scan with contrast is above the proposed reference level of 1050 mGy cm for head scans. Calzado et al [13] have shown that the correlation between image quality score and DLP is not straightforward for CT of the brain. In that study, higher DLP values were not always associated with greater fulfilment of some specific EC image quality criteria, for which fulfilment was uniformly low across all scanners tested. The highest score against these particular criteria was attributed to reconstruction algorithm rather than tohigher dose. Similarly, another criterion on twoscanners attracted low scores attributable to pathology, anatomical variants or motion artefact. Mean DLP values in the range 480–1018 mGy cm were reported, with the highest overall image quality score associated with the scanner giving a mean DLP of 1018 mGy cm. However, this scanner, despite having the highest image score and highest dose, gave the worst performance for two of the criteria. Because of these complex relationships, detailed analysis of our brain scans will be required to determine whether dose reduction without compromising diagnosticity is feasible.

Image noise is represented by the statistical variation of measured values of Hounsfield units when scanning a homogeneous object such as a Perspex phantom. A 2-fold increase in noise might be expected from a 4-fold decrease in dose, and this has been borne out to some extent in our standard chest examinations, where the dose has been halved, giving an observed increase of 90% in noise [4]. Decreases in dose were less dramatic for other scan protocols, and this is reflected in the smaller increases observed in noise levels.

The continual appraisal of images against the EC quality criteria during this first iteration in the optimization process has brought us to the stage where further dose reduction will require more detailed analysis of image quality, which will probably be on a smaller scale. There may be some scope for dose reduction in the future, arising from the use of alternative modalities, although currently the shift tends to be in the opposite direction, reflecting the national situation. CT scans for suspected pulmonary embolus and renal calculi are examples of such pathologies, formerly referred for examination by nuclear medicine and intravenous urography (IVU), respectively. We have effectively removed much of the excess dose caused by our CT practice, achieving a collective dose saving of 5.6 manSv per year, and leaving a smaller margin in which more detailed optimization of practice can be pursued.

The CT scanner at this site uses solid state detector technology. Zeman et al [14] have reported that the use of solid state detectors allows a theoretical dose reduction of approximately 30% in comparison with equivalent scanners with gas detectors. It follows that similar models of scanner with gas detectors might operate optimally at about 14 manSv for a similar patient workload. Similarly, extrapolating our average patient examination dose of 1.8 mSv per scan to an equivalent scanner with gas detectors, an average "optimized" examination dose of around 2.3 mSv might be expected.

With CT being such a major contributor to UK collective effective dose, it is crucially important that exposure from this source is justified for every patient, and once justified, that it is optimized. Manufacturers of CT scanners can assist in the process of optimization by further development of dose-saving design features, and by promoting and promulgating lower dose scan parameters during applications training after installation.

Received for publication July 14, 2000. Revision received January 12, 2001. Accepted for publication April 4, 2001.


   References
Top
 Abstract
 Introduction
 Method
 Results
 Discussion
 References
 

  1. National Radiological Protection Board. Patient dose reduction in diagnostic radiology. Documents of the NRPB 1990;1(3). Chilton: NRPB, 1990.
  2. Hughes JS. Ionising radiation exposure of the UK population—1999 review, NRPB Report 311. Chilton: NRPB, 1999.
  3. Crawley MT, Rogers AT. A comparison of CT practice in 1989 and 1991. Br J Radiol 1994;67:872–6.[Abstract]
  4. European Commission. EC quality criteria for computed tomography, EC Working Document, EUR 16262. Brussels: EU, 1998.
  5. Wade JP, Weyman JC, Goldstone KE. CT standard protocols are of limited value in assessing actual patient dose. Br J Radiol 1997;70:1146–51.[Abstract]
  6. Edyvean S, Lewis MA, Britten AJ, Carden JF, Howard GA, Sassi SA. Type testing of CT scanners: methods and methodology for assessing imaging performance and dosimetry, Evaluation Report MDA/98/25. London: Medical Devices Agency, 1998.
  7. Le Heron JC. CTDOSE. Christchurch, New Zealand: National Radiation Laboratory, 1993.
  8. Jones DG, Shrimpton PC. Normalised organ doses for X-ray CT calculated using Monte Carlo techniques, NRPB Software Report 250. Chilton: NRPB, 1993.
  9. Medical Devices Agency Evaluation Group. Imaging performance assessment of CT scanners (ImPACT). www.impactscan.org
  10. Shrimpton PC, Jones DG, Hillier MC, Wall BF, Le Heron JC, Faulkner K. Survey of CT practice in the UK Part 2: dosimetric aspects, NRPB Report 249. Chilton: NRPB, 1991.
  11. Clarke J, Cranley K, Robinson J, Smith PH, Workman A. Application of draft European Commission reference levels to a regional CT dose survey. Br J Radiol 2000;73:43–50.[Abstract]
  12. Korner E. Steering Group on Health Services Information, First Report to the Secretary of State. London: HMSO, 1982.
  13. Calzado A, Rodriguez R, Munoz A. Implementation of quality criteria for brain and lumbar spine CT. Br J Radiol 2000;73:384–95.[Abstract]
  14. Zeman RK, Fox SH, Silverman PM, Davros WJ, Carter LM, Griego D, et al. Helical (spiral) CT of the abdomen. AJR 1993;160:719–25.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 RadiologyHome page
J. Menke
Comparison of Different Body Size Parameters for Individual Dose Adaptation in Body CT of Adults
Radiology, August 1, 2005; 236(2): 565 - 571.
[Abstract] [Full Text] [PDF]

Home page
 Br. J. Radiol.Home page
L M Moran, R Rodriguez, A Calzado, A Turrero, A Arenas, A Cuevas, B Garcia-Castano, N Gomez, and P Moran
Image quality and dose evaluation in spiral chest CT examinations of patients with lung carcinoma
Br. J. Radiol., October 1, 2004; 77(922): 839 - 846.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
M. K. Kalra, M. M. Maher, T. L. Toth, L. M. Hamberg, M. A. Blake, J.-A. Shepard, and S. Saini
Strategies for CT Radiation Dose Optimization
Radiology, March 1, 2004; 230(3): 619 - 628.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
M. K. Kalra, C. Wittram, M. M. Maher, A. Sharma, G. B. Avinash, K. Karau, T. L. Toth, E. Halpern, S. Saini, and J.-A. Shepard
Can Noise Reduction Filters Improve Low-Radiation-Dose Chest CT Images? Pilot Study
Radiology, July 1, 2003; 228(1): 257 - 264.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Crawley, M T
Right arrow Articles by Wainwright, A
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Crawley, M T
Right arrow Articles by Wainwright, A

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
BJR DMFR IMAGING  ALL BIR JOURNALS