| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Vancouver General Hospital, Vancouver, British Columbia, Canada; and 2 University of British Columbia, Vancouver, British Columbia, Canada
Correspondence and requests for reprints should be addressed to John R. Mayo, M.D., Department of Radiology, Vancouver General Hospital, 899 West 12th Avenue, Vancouver, BC, V5Z 1M9 Canada. E-mail: John.Mayo{at}vch.ca
ABSTRACT
Computed tomography (CT) examinations are increasingly used for clinical diagnostic and research purposes, as they provide in vivo anatomic information similar to that provided by gross anatomy. In conjunction with physiologic maneuvers or contrast media, CT may also provide in vivo physiologic information. Using calibrated acquisition protocols, accurate noninvasive measurements of tissue density, air volume, blood volume, and capillary perfusion can be performed. Serial CT scans can provide longitudinal measurements indicative of disease progression or regression, allowing noninvasive assessment of treatment effects. However, the X-ray radiation associated with CT has been associated with a small but significant increased risk of malignancy, which may be fatal. Large studies have detected this small risk, which appears to be related to the cumulative radiation dose of all previous exposures in a linear fashion. It has been shown that the risk from a given radiation exposure is greater in young people and females compared with older males. The combination of these two risk-enhancing factors, found in pregnant females, provides the greatest risk. Radiation risk decreases with increasing age for both men and women, asymptotically approaching zero. Radiation risk can be calculated using dose metrics provided on current CT scanners as outlined in this article. Ethically, given that radiation is associated with measurable risk, clinically indicated and research CT examinations must provide an increase in knowledge that has substantial benefit to the subject. This benefit should be related to the potential of saving of life or to the prevention or mitigation of serious disease.
Key Words: radiation dose • computed tomography • longitudinal studies • cancer risk
The invention and rapid development of computerized tomography (CT) is one of the major medical advances of our time. Current multidetector CT scanners can image the entire chest in 2 to 5 seconds, producing up to 1,000 slices, each composed of sub-millimeter isometric voxels. These high signal to noise ratio, large field of view images provide noninvasive anatomic evaluation of the chest with information content similar to that achievable at autopsy. In vivo physiologic functional information regarding pulmonary and systemic perfusion or gas transport can be obtained by acquiring CT images while administering intravenous contrast media or performing breathing maneuvers, respectively. Expert radiologic interpretation of these images can differentiate diseases that are indistinguishable on history and physical examination but demonstrate unique changes at the gross anatomic level. Using standardized and calibrated acquisition protocols; accurate noninvasive measurements of tissue density, air volume, blood volume, and capillary perfusion can be performed. After validation of surrogate outcome measures, serial CT scans can provide longitudinal measurements of disease progression and treatment effects.
These capacities have made CT an invaluable diagnostic and research tool, accounting for the explosive growth of CT examinations in the last 20 years. It is estimated that in 2006 more than 62 million CT scans were obtained in the United States, as compared with about 3 million in 1980 (1). Similar increased utilization has been reported in studies from the United Kingdom (2) and Canada (3). Research studies are also increasingly using serial CT examinations for the noninvasive evaluation of disease progression and treatment effects.
However, the increased utilization of CT comes with a price: increased population radiation exposure. In most cases, adding CT to diagnostic imaging algorithms substantially increases patient X-ray radiation exposure. For example, a chest CT examination (3–6 mSv) delivers 60 to 120 times more radiation dose compared with a postero-anterior (PA) chest radiograph acquired using film (
0.05 mSv) or 90 to 180 times that of the same view obtained using digital radiography (0.03 mSv). In addition, since CT is so available and easy to perform, it is liberally applied to exclude potentially serious but statistically unlikely diagnoses, often solely to reassure anxious patients and clinicians. In the chest, serial CT studies are widely employed to assess disease progression in chronic obstructive pulmonary disease (COPD), interstitial lung disease, and cystic fibrosis. Repeated CT scans are often employed to follow suspicious lung nodules in patients at risk for lung cancer. The net result of these actions is greatly increased CT utilization and population radiation dose. It is noted that in some cases CT replaces examinations with higher dose (e.g., bronchography for bronchiectasis, nuclear medicine ventilation perfusion scintigraphy followed by pulmonary angiography for suspected pulmonary embolism) while providing equivalent or superior diagnostic information, but these situations are in the minority.
The increased radiation dose of chest CT compared with the PA chest radiograph arises from two properties of the CT technique (4). First, unlike analog film radiography, in which the image acquisition and display are both reliant on the film, CT is a digital technique in which image acquisition and display can be independently manipulated. Therefore, when CT dose is excessive, the image does not become too dark (as it does in film radiography), but instead improves because of decreased image noise (5). Second, visualization of image noise is enhanced by the ability to map the entire visible gray scale onto a selected segment of the CT number scale. As a result, image degradation due to quantum noise (mottle) is easily visible and interferes with image interpretation. At high noise levels, images may be clearly nondiagnostic. However, at lower noise levels more subtle image degradation occurs which may lead to diagnostic inaccuracies or a lack of confidence in image interpretation, effects that are difficult to detect and measure. Image noise also affects the accuracy of measurements made on chest CT images (e.g., assessment of the extent of emphysema). Failure to standardize acquisition protocols and account for equipment specific differences in noise levels can lead to systematic errors in measurements that may introduce errors in surrogate measures of disease activity and progression.
In the absence of standardized validated protocols, radiologists often obtain CT images using high radiation exposure levels to minimize image noise and maximize image quality. Studies in multiple jurisdictions have shown that the lack of standardization leads to wide variation in the level of radiation administered for the same CT examination between institutions (2, 3), with no detectable difference in patient outcomes. In addition, studies have shown that radiologists, referring clinicians, and patients may be unaware of the high level of radiation exposure associated with CT examinations (6). This knowledge gap undoubtedly contributes to the overuse of CT in low-yield diagnostic situations and its overuse in following disease progression or treatment effects.
In the early 1990s, concern was raised regarding radiation dose in chest CT (7–9). The authors of these early studies suggested that greater consideration needed to be given to optimizing CT exposures and ensuring appropriate clinical use guidelines on the single detector row CT scanners in use at that time. These early warnings were not heeded, and the last 15 years have been characterized by ever-increasing CT utilization. The increased utilization of CT has been hastened by the development of multidetector row CT scanners, leading to expanded clinical indications for CT examinations (e.g., pulmonary embolism, cardiac gated CT angiograms [CCTA], trauma CT). This has fueled an increase in both the number of installed scanners and the number of patients scanned per shift. These advances have served to further increase population CT radiation exposure. The development of evidence-based guidelines governing the use and technical parameters for CT is required to responsibly use this valuable diagnostic test.
The purpose of this review is to outline (1) evidence indicating the detrimental effect of radiation dose at the level administered in chest CT examinations, (2) parameters that affect CT radiation dose, (3) advances in dose reduction in the chest CT, and (4) the interaction between CT radiation dose and diagnostic accuracy. A complete review of radiation dosimetry and bioeffects is beyond the scope of this review.
RADIATION BIOEFFECTS
There has been considerable debate within the medical community regarding the risk of low-level radiation exposure from CT. The reason for this debate arises from an incomplete knowledge of the complex link between ionizing radiation and future negative outcomes in humans. In broad overview, the negative outcomes of ionizing radiation in humans can be divided into two major categories that can be separated on the basis of time and exposure: deterministic effects seen immediately after large exposures, and stochastic effects seen after a long latent period (6–25 yr) and associated with low exposures.
Deterministic effects, skin erythema, skin necrosis, and hair loss only occur above a threshold dose that lies well above those administered in diagnostic chest CT examinations. In medical imaging, deterministic dose levels are only seen in complex interventional cases using large quantities of fluoroscopy time. These effects will not be discussed further in this review.
By comparison, stochastic effects are believed to have no radiation dose threshold, and therefore are associated with the low radiation doses delivered during chest CT. Mechanistically, stochastic effects are believed to be mediated by chemical damage to the DNA molecule and clinically manifest as an increased risk of cancer and genetic defects. Stochastic effects occur randomly and the risk of their occurrence depends on the type of ionizing radiation administered, the tissue receiving the radiation, and the age of the subject. It is believed that dose fractionation, a substantial modifier of detrimental effect for deterministic radiation doses, does not substantially modify the stochastic risk (10). Stochastic risks are believed to be cumulative, with increasing risk seen over successive exposures.
Subjects exposed to the atomic bomb explosions in 1945 have been extensively studied in the last 60 years. This group is unique since it is large, covers all ages, and was not selected on the basis of underlying disease. A substantial portion of the survivors received less than 50 mSv, a low level of exposure that approximates the dose range delivered by multiple chest CT exams. The major negative effect seen in this group is an increase in the number of cancers over that found in a nonexposed population. An earlier presentation of cancers has not been observed. However, to assess the risk from a single chest CT scan requires extrapolation of these results to even lower doses, and the nature of this extrapolation has proven to be highly controversial.
Disagreement regarding the extrapolation of nuclear explosion data is based on three nonresolvable issues: uncertainty in the actual radiation exposure received, since on-site radiation dose measurements were not obtained; differences in the natural cancer risk of the Japanese population compared with other populations; and the different quality of the radiation imparted by atomic bombs compared with X-ray–based medical imaging. As a result of differences in interpretation, learned societies have come to varying conclusions on the risk attributable to radiation exposure at the levels found in chest CT. The International Commission on Radiological Protection, or ICRP, used a linear no-threshold extrapolation of nuclear explosion data and estimated 50 additional fatal cancers induced per million people exposed to 1 mSv of medical radiation (11). In contrast, the French Academy of Science concluded that there was not sufficient evidence to support an increased cancer risk associated with radiation exposures less than 20 mSv (12), a level above that delivered in chest CT examinations (< 6–11 mSv). Further conflicting evidence on the impact of low-level radiation exposure is found in tissue culture experimental studies that have shown induction of free radical detoxification mechanisms with low-level radiation exposure (13). This has led some to suggest that low-level radiation exposure may be beneficial, an effect known as radiation hermesis. Finally, the long-term study of the mortality of British radiologists showed lower cancer mortality than predicted by the atomic bomb data (14). It is postulated that this may be accounted for by the healthy worker effect, the beneficial effects of dose fractionation, or overestimation of the dose received by these physicians.
In 2007, additional important data were added to this debate (15) when the 15-country study reported the cancer induction effect of low-level radiation exposure studied in 407,000 radiation workers followed for over 20 years providing 5.2 million person-years of follow-up. This study is unique as it reports on the largest cohort to date, has accurate dosimetry, and investigated multiethnic workers. Ninety percent of the subjects received a dose less than 50 mSv and on average each worker received a dose of 19 mSv. Therefore this study is focused on low-level doses, close to that received during a single chest CT examination (6–11 mSv). The authors reported an excess relative risk (ERR) for all-cause mortality of 0.42 per Sievert (0.00042 per mSv), with a statistically significant increasing excess relative risk with increasing radiation dose (P < 0.02) indicating a dose–response effect. The increased risk in all-cause mortality was mainly due to an increase in mortality from all cancers.
A subanalysis stratified by dose categories (less than: 400, 200, 150, and 100 mSv) showed that cancers in the highest dose categories did not drive the risk estimates. Therefore, this study supports the concept that there is a small cancer risk from low-dose radiation delivered in CT examinations. These new data add supportive evidence to the concern over radiation dose delivered in chest CT examinations and support the use of the ALARA principle (As Low As Reasonably Achievable) for these exams.
However, there are limitations to these new data. Because workers were studied, there is no information on the effect in children; and since 90% of the workers were men who received over 98% of the cumulative dose, minimal information is available on the effect in women. The largest excess mortality from all contributing countries is found in the data from Canada, and statistical significance is lost if this cohort is not included. Finally, the largest discrepancy between this study and the atomic bomb cohort arises in the lung cancer mortality, suggesting that the confounding effects of smoking may have been inadequately allowed for.
The influence of age at exposure and of sex has been studied in the nuclear explosion cohort, showing that radiation risk is substantially modified by these subject factors (16, 17). The increased radiation sensitivity of children is felt to arise from two biological facts: they have more time to express the cancer-inducing effect of radiation and they have more rapidly dividing cells than adults which are inherently more radiation sensitive.
It has been found that women have approximately twice the risk compared with males for the same level of radiation exposure. Increased female risk is heightened in chest CT by the presence of radiosensitive breast tissue in the radiated field. Radiation dose to breast tissue in chest CT examinations has been calculated (18) and directly measured (19, 20), with reports showing wide variation in average values, ranging from 10 to 70 mGy. The variation in values is related to CT parameter settings, differences in size and configuration of breast tissue, and methods to calculate or directly measure radiation dose. There is no debate that all CT-associated breast radiation dose values are substantially greater than the average glandular dose of 3 mGy for standard two-view screening mammography. It is important to note there is a strong age at exposure effect for breast tissue, with lower risk for subjects above the age of 40 (21). These factors must be taken into account in setting chest CT radiation dose parameters in CT chest examinations for women. Breast shields, thyroid shields (22, 23), and X-ray tube current modulation techniques have been employed to decrease radiation dose to these superficial and radiosensitive tissues within the chest. These techniques have been shown to decrease breast radiation exposure delivered in chest CT scans. However, these dose-modifying techniques must be used with consideration of their impact on image quality.
RADIATION DOSE MEASUREMENT
There are many methods currently in use for quantifying ionizing radiations (Table 1) (24). The fact that several methods exist attests to the complexity of this issue. The simplest parameter, radiation exposure, is determined by measuring ionization in air caused by the X-ray beam. The measurement unit is coulombs per kilogram (abbreviation, C/kg). It has limited clinical value, as it does not take into account the area irradiated, the penetrating power of the radiation, or the radiation sensitivity of the irradiated organs. From radiation exposure we can calculate the skin entrance dose, which is important when examining deterministic effects such as skin erythema. Although deterministic effects are not encountered in routine CT, they are of potential concern in CT fluoroscopy. A more refined measurement is absorbed dose, determined by measuring the energy absorbed per unit mass within an object. The measurement unit is the gray (abbreviation, Gy). Unlike radiation exposure, the gray is dependent on the composition of the object or subject placed in the radiation beam. However, absorbed dose does not account for the differing radiation sensitivity of organs, and it cannot provide a whole-body risk estimate or be used to facilitate comparisons between examinations in different parts of the body. Equivalent dose is a modification of absorbed dose that incorporates weighting factors to account for the different biologic effect of various sources of radiation. For X-rays, the radiation weighting factor is 1 and the equivalent dose has the same numerical value as absorbed dose.
|
|
|
Radiation dose surveys have noted wide variations in DLP settings for identical examinations between institutions (2, 3, 26, 27). To decrease this variation and protect the public from inadvertent overexposure, the European communities have published suggested reference dose values (28) for chest CT examinations, with a DLP value of 650 mGy cm. This reference dose value was obtained by surveying a large number of institutions in Europe and adopting the 75th percentile of responses as the reference dose values. This value serves as a guide to acceptable practice.
The European Commission has also provided guidelines for radiation exposure in medical and biomedical research (29). These serve as a guide to those planning or evaluating research (e.g., ethics committees). Research radiation exposure is divided into four categories (I, IIA, IIB, and III) corresponding to effective dose limits of less than 0.1 mSv, 0.1 to 1mSv, 1 to 10 mSv, and over 10 mSv. The increasing dose limits in the four categories are related to increasing potential benefit from the research as indicated by the category guidance notes: I, expected to only increase knowledge; IIA, increase in knowledge leading to a health benefit; IIB, increase in knowledge aimed directly at the diagnosis, cure, or prevention of disease; III, increase in knowledge to have substantial benefit and usually directly related to the saving of life or the prevention or mitigation of serious disease. In addition, these research guidelines account for the substantial age variation in radiation sensitivity allowing an increase by a factor of 5 to 10 for subjects over 50 years old and decreasing the limits by a factor of 2 to 3 for children. It is noted that research that involves serial radiologic investigations must calculate the cumulative radiation dose over the course of the study to determine the exposure category.
RADIATION DOSE REDUCTION
Reduction in CT radiation exposure results in increased image noise and decreased image quality. Studies assessing the subjective evaluation of chest CT scans have demonstrated that radiologists consistently gave higher image-quality scores to images obtained with a higher radiation dose (30, 31). Image noise can be measured by placing a region of interest (> 100 pixels) in an area of uniform density (e.g., the thoracic aorta). The standard deviation of the pixel values represents image noise and is a measure of the uncertainty of quantitative CT measures. It is noted that the choice of reconstruction algorithm affects image noise, with higher noise associated with high-spatial-frequency reconstruction algorithms (e.g., bone or lung algorithms) compared with low-spatial-frequency algorithms (e.g., standard, soft-tissue algorithm). Since high-spatial-frequency reconstruction algorithms are most commonly used to assess bones or lung parenchyma, tissues with high radiographic contrast, increased noise is usually not a diagnostic problem. However, increased noise may interfere with quantitative measures of disease such as computer-calculated emphysema scores. New adaptive reconstruction algorithms are being developed that can decrease image noise, providing improved image quality at lower radiation dose. These advanced algorithms should facilitate radiation dose reduction.
Radiation dose can be adjusted at the time of image acquisition by changing the X-ray tube current or voltage and the scan time. In practice, the tube current is most frequently adjusted to change the radiation dose and image noise. In most CT scanners, the tube current is adjustable in steps from 20 mA to approximately 400 mA. Decreasing the tube voltage also decreases the radiation dose, but also affects subject contrast and can impact CT number measurements. Finally, radiation dose is linearly related to the scan time. However, in most cases scan time is minimized to reduce motion artifact. It is noted that the radiation exposure delivered at a given tube voltage and current setting will vary greatly between CT scanners of different models and manufacturers because of differences in scanner geometry (X-ray tube-to-patient separation) and X-ray tube filtration.
In the past, the tube current of CT scanners was uniform at all angles around the patient and for the full longitudinal (cranial caudal) extent of the scan. However, the chest is an elliptical object that has higher attenuation from left to right than from anterior to posterior. Attenuation also varies as the chest is scanned cranial to caudal because of the shoulders. CT image quality is disproportionately degraded by views with few photons (photon starvation) compared with the image quality improvement associated with views with high photon counts. To address this issue, manufacturers have introduced programs that adjust the tube current depending on the attenuation of the object in both the transverse (x, y) and longitudinal (z) directions to minimize either photon-starved or photon-rich projections, maximizing image quality while minimizing radiation dose. This tube current modulation technique has been shown to produce a substantial reduction in radiation dose (32–34) with minimal degradation of image quality. Routine use of dose modulation systems is recommended, as they compensate for asymmetry in the size and density of the body section being scanned, resulting in a signal-to-noise ratio that is adequate for diagnosis but is not excessive (35). Advanced tube current modulation schemes with novel reconstruction algorithms are being developed to reduce radiation dose to superficial radiation sensitive-tissues such as the breast and thyroid. Further experience with these new radiation dose modulation systems is required before they can be widely employed.
However, dose modulation systems can produce variations in image noise that may modify the numeric evaluation of emphysema in serial follow-up examinations. In addition, dose modulation systems may interact with the patient position and X-ray beam filters, producing increased radiation dose in patients who are incorrectly centered in the CT gantry. Algorithms to automatically center patients are being developed (36). In longitudinal research studies careful attention to scan parameters and patient positioning is an important component of both radiation dose reduction and inter-scan reproducibility. Finally, repeated scanning of the same region increases the radiation dose in a linear fashion. Therefore, the timing of follow-up examinations involves a trade off between additional information and the radiation dose detriment of the associated dose buildup effect.
CONCLUSIONS
The introduction of helical and multi–detector row CT scanners has resulted in an increase in the number of indications for and the diagnostic accuracy of chest CT examinations. However, despite increasing education and awareness, the current level of radiation exposure from CT remains high. The 15-country study has added further information to support the linear, no-threshold approach to the detrimental effects of radiation dose in the range below 20 mSv, similar to that received in chest CT scans (3–6 mSv). Current radiation dose surveys continue to indicate that there is large variation in the technical factors employed by radiologists (3), and there is a resultant large variation in the radiation dose to patients between institutions. Reference dose values for chest CT have been developed and published. Radiologists need to monitor the radiation dose delivered in examinations within their institutions, adopt and adhere to national radiation dose guidelines, and investigate further dose reduction strategies within their own practices. Further research into the complex relationship between radiation exposure, image noise, and diagnostic accuracy should be encouraged to scientifically establish the minimum radiation doses that provide adequate diagnostic information for standard clinical questions, disease quantitation, and disease follow up. Once these minimum levels of image quality are determined and validated, automatic exposure controls for CT scanners should be programmed to ensure that all patients undergo CT with techniques that conform to the ALARA (As Low As Reasonably Achievable) principle. Finally, new approaches to image reconstruction should be addressed to maximize the relationship between dose and image quality.
As dispensers of this known carcinogen, radiologists must take the lead in promoting all of these measures for patient protection. Since children, young adults, women, and pregnancy have been shown to increase radiation sensitivity, the most strident dose reduction efforts should be focused on these groups. Finally, it is noted that the complexity of CT requires a close collaboration between radiologists and medical physicists to successfully reduce radiation dose while maintaining diagnostic accuracy.
FOOTNOTES
Conflict of Interest Statement: J.R.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
(Received in original form August 4, 2008; accepted in final form September 23, 2008)
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
