This Article Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Martin, C J Search for Related Content PubMed PubMed Citation Articles by Martin, C J

20 years of patient dose studies: where should we go from here?

Department of Clinical Physics and Bioengineering, Gartnavel Royal Hospital, Health Physics, West House (Ground Floor), 1055 Great Western Road, Glasgow G12 0XH, UK

The first comprehensive survey of patient doses for a range of diagnostic X-ray examinations was published in the UK in 1986 by the National Radiological Protection Board (NRPB) [1]. Concern about the wide variation in patient doses for similar examinations revealed in the survey persuaded medical physicists and others to measure doses in their own hospitals and investigate whether changes in equipment or technique should be made when these were found to be high. The last 10–15 years has seen a regular quota of patient dose studies for radiology examinations published in the literature. There is, on average, one in each issue of the British Journal of Radiology at the present time. The NRPB and others have derived tables of conversion coefficients to allow effective doses to be calculated from the various dose quantities normally measured so that doses from different examinations can be compared [2]. Estimates of potential risk from developing cancer can be derived using coefficients published by the International Commission on Radiological Protection (ICRP) [3]. Much useful work has been done and there are always more things that can be investigated. However, there comes a point where the benefit is limited and more may be achieved through application of effort in other areas. We have to ask questions about the need for further studies of patient dose and the level of quantification that is justified. What should we be trying to achieve from a patient dose study? When is it appropriate to publish results from such studies? Ought we to pay more attention to the impact on image quality, and if so, how do we assess it? How far do we go in our dose calculations? Are we justified in drawing conclusions about effective dose and risk, since factors that we use in the calculations change as the state of knowledge develops?

One of the main benefits from patient dose surveys has been to provide information on doses in different hospitals for comparison, which give an indication of whether the techniques used in our own hospitals are optimized in terms of dose. The introduction of diagnostic reference levels (DRLs) through the Ionising Radiation Regulations [4] has formalized this process in the UK. The accumulation of patient dose data on which DRLs can be based is important. The NRPB (now part of the Health Protection Agency) has collated doses from hospitals throughout the UK and publishes reports presenting these data every 5 years [5]. These reports provide us with a large databank of doses for a wide range of hospitals that we can draw on, and are ideal for setting national DRLs.

The time has passed when results of a patient dose survey from a particular region of the country should be published just to provide information on dose magnitudes, but independent studies can make important contributions to our knowledge. They can show how different techniques influence dose, because radiology staff and physicists can analyse techniques and identify factors contributing to larger doses, and this information is useful in optimization. However, we should review critically the aims and objectives before undertaking a study. The following are examples of the questions we might ask ourselves.

• Is the procedure to be studied representative of ones that are performed widely?
  
• Are there sufficient results from other studies on current dose levels for the examination(s)?
  
• Will the study provide new data on general dose levels in a particular country?
  
• Are the data going to assist others in setting DRLs?
  
• Will the study provide information on relative merits of different techniques?
  
• Are results likely to be useful in optimization?
  
• Does the study contribute new ideas on methods for dose measurement?
  

If simple studies are to present dose information that is representative of current practice, results from at least 3–4 hospitals are required and preferably 8–10. Doses for fluoroscopy equipment will be strongly influenced by the equipment performance and facilities available, as well as the radiologists' techniques, so studies on one or two units may give misleading results. If there are only one or two hospitals carrying out the procedure in an area, then a joint study with colleagues from other centres might be more appropriate. Otherwise, probably the best course of action for UK groups is to send results to the NRPB for inclusion in their next patient dose report. Some of the examinations for which there is less information available are in interventional radiology and cardiology. A problem with studies on complex procedures such as these is the difficulty in defining standard procedures, since the vessels investigated have a major influence on the dose. Categories of procedures must be worked out carefully so that like is compared with like.

So far we have considered patient dose in isolation, but this is only half of the optimization equation [6]. The most important component is ensuring that the image quality is good enough to enable the diagnosis to be made. We cannot reduce dose indefinitely without affecting image quality. The consequences of a failure to make a correct diagnosis because of poor image quality are immediate and the risks are far greater than those from the radiation exposure. There are some applications where the relative positions of structures are being assessed, such as bone alignments, where a noisy image obtained with a low dose of radiation will give all the diagnostic information required. However, for the majority of applications we need as high quality an image as the technique can provide, since it is the fine detail which allows a definitive diagnosis to be made. Some thoughts about assessment of image quality are given below.

• We should look at simple contrast to noise ratios to determine the best exposure parameters to use for imaging different parts of the body. The contrast to noise ratio is the signal to noise ratio (SNR) which can be measured for larger structures with areas covering hundreds of pixels.
  
• We should look at SNR as a function of spatial frequency (detail size) to determine how well our system can visualize structures of different size. This can be accomplished in terms of Detective Quantum Efficiency or Noise Equivalent Quanta, which provide objective assessments of imaging performance for an ideal observer.
  
• We should establish methods of image quality measurement that are linked more closely to the perception tasks involved in making diagnoses.
  
• We need to establish links between the sizes and attenuating properties of structures important in clinical diagnosis for different organs and tissues, and details in test objects that might be included in imaging phantoms.
  
• We should make use of computer simulations to evaluate imaging performance and predict optimum exposure conditions for different examinations.
  
• We need to understand the effects of image processing techniques on the appearance and perception of digital images.
  
• More ready access to data from digital imaging equipment will allow image analysis and information on the look up tables used in producing the displayed image would be advantageous.
  

The final question to consider in relation to patient dose studies is what information should be provided on dose? The measured quantities dose–area product, entrance surface dose, CT dose index and dose–length product are the most important as they can be used for direct comparisons. However, it has been the practice to provide information on assessment of effective dose as well. There are large uncertainties in the coefficients used to convert dose measurements to effective dose. They are based on risk coefficients for a cross-section of the population with all ages and both genders, which is clearly not applicable to the population undergoing diagnostic medical examinations. The ICRP have suggested in a draft of their future recommendations that effective dose should only be used for establishment of prospective radiation protection guidance, not for assessing risks of stochastic effects at low doses in retrospective situations. Nevertheless, effective dose is useful as a comparator for relative magnitudes of doses for different types of examination, but we must not attribute to this more certainty than is justified. This is especially true since the definition of effective dose published by ICRP will change as more information becomes available from epidemiological studies. For examinations where particular radiosensitive organs receive significant doses, it may be more appropriate to give estimates of doses to these organs in the assessment, rather than effective dose. For complex examinations, detailed analyses of dose for each projection have been undertaken. This has allowed an indication of effective dose to be obtained for particular examinations, which has been interesting. However, this level of detail requires a significant effort to provide information that is of limited value. A simple presentation of measurable dose quantities with advice on techniques may be more appropriate in future studies.

While dealing with patient dose assessments, it is worth considering what information should be given as an indication of potential harm to an individual who as a patient has received an overexposure or who is to take part in a research study. Any risk assessment is only indicative, because, not only is the dose calculation an approximation, but also the accuracy of the risk assessment depends on the validity of underlying assumptions used in making extrapolations of limited data from epidemiological studies. Placing a numerical value on the risk for a specific case suggests a degree of certainty that is not justified. The ICRP indicate in a draft of their revised recommendations that risk coefficients are intended for assessment of risks in general terms, not for retrospective assessment of risks in specific populations. In fact risk assessment calculations are based on the linear no threshold (LNT) dose–effect model, which may represent a worst case. Use of general terminology is more appropriate for conveying our state of knowledge [7]. Bearing in mind doses received from other sources such as natural background radiation (2.3 mSv per year on average in the UK), the risks derived from the LNT model, and the fact that the exact value of the excess risk of developing cancer at doses below 200 mSv is a matter of debate [8, 9], it is more appropriate to use general terms for risks associated with the low radiation doses received from diagnostic procedures. The following are offered as suggestions for describing the risks associated with effective doses in different ranges.

• <1 mSv negligible/trivial
  
• 1–10 mSv very low
  
• 10–100 mSv low
  

More explanation of the risk is appropriate for doses above 100 mSv.

In conclusion, a significant effort has been put into determining dose levels for radiology examinations over the last 20 years and much useful information has been obtained. We now need to be more selective in the studies that we choose to undertake and think carefully about what we want to achieve from each. Effective dose is an ephemeral quantity that evolves with time. Too much effort should not be wasted on complex calculations to derive a quantity which may be of limited value. The way forward should be through optimization involving all aspects of the imaging process in which measurement of dose is only one component. Finding the appropriate level of image quality is the most important objective. Keeping the dose low should always be secondary.

Received for publication February 2, 2005. Revision received February 15, 2005. Accepted for publication February 15, 2005.

References

1. Shrimpton PC, Wall BF, Jones DG, Fisher AS, Hillier MC, Kendall GA, et al. A national survey of doses to patients undergoing a selection of routine X-ray examinations in English hospitals. NRPB-R200. Chilton: NRPB, 1986.
2. Hart D, Jones DG, Wall BF. Estimation of effective dose in diagnostic radiology from entrance dose and dose area product measurement. NRPB-R262. Chilton: NRPB, 1994.
3. International Commission on Radiological Protection. 1990 recommendations of the International Commission on Radiological Protection. ICRP Publication 60. Ann ICRP 1991;21(1–3).
4. The Ionising Radiation (Medical Exposure) Regulations. SI (2000) 1059. London: HMSO, 2000.
5. Hart D, Hillier MC, Wall BF. Doses to patients from medical X-ray examinations in the UK – 2000 Review. NRPB-W14. Chilton: NRPB, 2002.
6. Martin CJ, Sutton DG, Sharp PF. Balancing patient dose and image quality. App Radiat Isot 1999;50:1–19.
7. Harrison JR. Health of the nation. Radiation Protection Bulletin No. 184, pp 10–13. Chilton: NRPB, 1996.
8. United Nations Scientific Committee on the Effects of Atomic Radiation. Sources and effects of ionising radiation, Vol. II Effects. New York: United Nations, 2000.
9. Kellerer AB. Radiation risk – historical perspective and issues. J Radiol Prot 2002;22:A1–A10.[CrossRef]

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]

				BJR Review of the Year - 2005. Br. J. Radiol., March 1, 2006; 79(939): 183 - 187. [Full Text] [PDF]

This Article Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Martin, C J Search for Related Content PubMed PubMed Citation Articles by Martin, C J