Advances in multidetector CT and post-processing have led to increasing utilization of CT cross-sectional imaging to approximately 88 million scans in 2010, an over 20-fold increase since 1980 [1]. This increase has resulted in an ~600% increase in U.S. per capita exposure to ionizing radiation from medical procedures from 1980 to 2006, with CT imaging contributing to nearly 50% of this increase [2]. During this time, the increasing temporal and spatial resolution of CT, as well as the ability to individualize and decrease radiation dose, have led to an increasing number of specialized CT exams for screening and organ-specific imaging [3–8]. While familiar with the hyperbole of sensationalistic media reports about the risks of radiation from CT, we have noticed that the lack of discussion between patients, radiologists, and referring physicians has led to fear of CT imaging on the part of many patients—leading them to forgo CT imaging all together—and eliminating the potential benefit CT may provide for their medical care. For example, we have observed physicians unwilling to undergo CT themselves even though alternative MR imaging is less accurate for their suspected diagnosis; patients with hepatic malignancies worried about the increased risk of death from repeat CT examinations; and symptomatic Crohn’s patients who are hesitant to obtain a CT enterography for the assessment of disease severity and complications. While it may be scientifically accurate to say that a typical abdominopelvic CT examination requires 400 times the dose of a chest X-ray [9], a chest X-ray cannot diagnose Crohn’s disease or metastatic colon cancer, providing information that would likely lead to beneficial therapy for the patient (Fig. 1). While CT radiation dose has fallen over the past three decades [3], new technical innovations such as automatic exposure control, automatic tube potential selection, iterative reconstruction, and other noise reduction techniques provide a new opportunity for radiologists to re-engage patients and clinicians in the discussion of the benefits and risks of CT.

Fig. 1
figure 1

35-year-old female symptomatic Crohn's patient with fear of radiation-induced cancer from CT imaging. Despite extreme difficulty with eating solid foods, patient was unwilling to undergo CT enterography and was waiting weeks for MR enterography examination under general anesthesia (due to claustrophobia). After benefits and risks of enterography were explained, patient underwent CT enterography exam the next day, demonstrating multifocal, long-segment jejunal (white arrows-A and C) and ileal inflammation (open arrows-B and C), facilitating treatment with resolution of symptoms.

Full size image

Risk estimates, uncertainty, and recommendations

CT volume dose index (CTDIvol) is a measure of the radiation delivered by a CT scanner for a given exam [10, 11]. It reflects the mean dose to the center of the scan volume to a standardized acrylic cylinder, and it is automatically reported by CT scanners. It tells users what amount of radiation the scanner is producing, but other factors need to be taken into consideration to estimate the dose absorbed by the patient [12]. The CTDIvol for each CT exam is automatically recorded in the patient’s imaging record. While this information is useful for radiologists and physicists monitoring and comparing doses between acquisition protocols, patients and their referring clinicians want to know the biologic risk of CT examinations, not a measure of scanner radiation output. Biologic risk is conveyed by effective dose, which normalizes partial body radiation exposure to a whole body exposure based on equivalent risk. To calculate effective dose, organ doses must first be estimated. Organ-specific doses are then multiplied by weighting factors (which change depending on expert consensus at the International Commission on Radiation Protection) and then summed to calculate the effective dose. This weighted sum is a broad measure of biologic risk for radiation protection purposes, but “cannot be measured and cannot be used for individual risk assessment” [1, 13, 14]. Effective dose is useful for exam optimization, comparison between exams, and Institutional Review Boards, but does not convey the risk for any particular patient. This is particularly true since the weighting factors used to calculate effective dose are an average value over both genders and all ages. In addition, individual patients have differing baseline risks [14, 15].

The National Academy of Sciences report on the Biological Effects of Ionizing Radiation VII (BEIR VII) evaluated the biologic impact of low doses of radiation based on a combination of data from atomic bomb survivors, medical and environmental radiation studies, and occupationally exposed workers [16]. They concluded that the available evidence is consistent with a linear non-threshold (LNT) risk model for cancer risk, although the LNT model was not statistically significantly better than nonlinear or threshold-based models at doses in the range of diagnostic imaging exposures. The atomic bomb survivor data demonstrate increases in excess relative risk of cancer for acute, high dose-rate exposures having effective doses above 150–200 mSv, but there is considerable disagreement on the excess relative risk of cancer in the dose range below this level [17–21]. The 15-country workers study conducted by the International Agency for Research on Cancer (in over 407,000 nuclear workers with an average cumulative dose of 19 mSv) reported a small but significant excess relative risk estimates for all cancers, excluding leukemia [22]. However, this result has since been discredited due to errors identified in the Canadian data, which accounted for the majority of cancers. In June 2011, the Canadian Nuclear Safety Commission issued a report noting no significant increase in cancer risk in radiation workers [23]. Muirhead examined 174,000 British radiation workers and found that for leukemias (excluding CLL) and solid neoplasms, confidence intervals did not exclude the absence of risk until exposures were over 200 mSv [24].

Cumulative doses above 200 mSv can be reached in patients with a long history of medical imaging. Fractionation is a process whereby a larger cumulative dose of ionizing radiation is given in smaller increments. Howe [25] studied a cohort of over 64,000 Canadian tuberculosis patients, of whom 39% were exposed to highly fractionated multiple chest fluoroscopies leading to a mean lung radiation dose of 1020 mSv, and observed no significant increases in cancer risk. He concluded there is “strong support…for a substantial fractionation/dose-rate effect,” although the magnitude of this effect may vary between different target organs [26].

Multiple organizations and reports have given recommendations for describing biologic risks to patients from medical radiation (Table 1). The American Association of Physicists in Medicine has said that predictions of hypothetical cancers and cancer deaths are highly speculative, and the Health Physics Society has recommended against a quantitative estimation of health risks below effective doses of 50 mSv [27, 28]. While the BEIR VII report notes that statistical limitations make it difficult to evaluate cancer risks in humans less than 100 mSv, it also states that it is “unlikely that a threshold exists for the induction of cancers…[and] that the occurrence of radiation-induced cancers at low doses will be small” [29].

Table 1 Consensus statements regarding estimating the describing biologic risks to patients from medical radiation
Full size table

We have found that it is extremely useful in discussing the biologic risk of CT examinations to refer to relative risks for which the patients may be more familiar (Fig. 2). For example, the overall probability of dying from cancer is 23%. Using an estimated dose of 10 mSv for an abdominopelvic CT scan and the linear no-threshold model, a conservative estimate of the increased probability of dying from cancer after an abdominal pelvic CT would be 0.05%. In comparison, the increase in risk from the radon in U.S. homes is estimated to be 0.3% (six times higher), and the estimated increase in risk from arsenic in drinking water in the average U.S. home is 0.1% (twice as high) [30]. Such comparisons are particularly helpful if used to engage a patient in thinking about potential benefit and potential risk. For example, the risk of dying from drowning is 0.09%, and the risk of dying from bicycling is 0.02%, and patients often think about swimming or biking as safe and derive benefit from participating in these activities. Extending this benefit-risk analogy, if a CT might potentially increase their survival by leading to potentially life-saving therapy by even a mere 1%, the CT would be 20 times more likely to be beneficial than harmful.

Fig. 2
figure 2

Visual representation of probability of death from various causes, compared to dying from a radiation-induced malignancy from abdominal or head CT, using risk assumptions as delineated in BEIR VII and linear no-threshold hypothesis. Sources are www.cdc.gov/nchs/fastats/deaths.htm and Gerber et al. [30].

Full size image

Net benefit

CT exams must be justified so that there is a net benefit to the patient, i.e., the exam is medically indicated and benefits exceed the risks. Physicians often weigh the benefits vs. risks for many types of drugs. In instances where there is a small risk, such as CT imaging, the quantification of the net benefit is driven largely by the perceived benefit. When the potential benefit for CT scanning is lower, such as in a screening study in asymptomatic patients, it is desirable to use lower radiation doses and no intravenous contrast in order to minimize risk (e.g., screening CT colonography (CTC) or low dose lung cancer screening). Berrington de González et al. [31] examined the benefit to risk ratio in CTC, assuming the radiation dose levels used in the National Colonography Study and colonographic screening every 5 years from ages 50–80, along with three microsimulation models for colorectal cancer development, and compared potential lives saved using screening CTC to potential deaths from fatal cancers due to medical radiation. They estimated a large benefit to risk ratio for screening CTC, which varied from 24:1 to 35:1.

The potential net benefit from any particular CT exam rests largely on patient risk for a particular disease, the potential impact of medical or surgical therapy on that disease, and the ability of CT to detect the relevant abnormality. All three of these factors must be taken into account when thinking about potential benefit. For example, patients who have a condition that CT can accurately detect and stage, and/or patients who are symptomatic or in whom the likelihood of disease is high, are likely to derive a high degree of benefit from CT imaging. In screening cohorts, patients are asymptomatic, and the potential benefit may be less due to lower disease prevalence (e.g., about 15% of screening CTC patients will have a colorectal polyp ≥6 mm). Benefits of CT compared to alternative imaging strategies must also be considered, in addition to the risk of not obtaining a CT exam. If multi-organ imaging is necessary (for example, in the trauma patient), CT imaging offers a clear advantage to alternative MR and ultrasound imaging because of its ability to image multiple organs and large portions of the body in a rapid period of time, a factor which has been cited by referring clinicians as advantageous to CT [32]. The availability of CT, ultrasound, or MR imaging and local radiologic expertise is also important, especially if conditions need to be ruled in or out quickly [32]. Comorbidities or idiosyncratic patient factors (e.g., renal failure, claustrophobia, metallic implants, inability to hold still, and compliance with screening recommendations) are all important patient-specific factors that clinicians might consider. Table 2 lists factors that clinicians and radiologists should consider when thinking about the potential benefit and risks for a CT examination.

Table 2 Factors that clinicians and radiologists should consider when thinking about the potential benefit and risks for a CT examination
Full size table

On the risk side of the equation, the appropriateness of CT imaging must be compared to potential alternatives, especially for younger or pregnant patients, where the potential risk of radiation-induced malignancy is higher [1, 33, 34]. Alternative ultrasound or MR imaging may be more justified if these imaging techniques can be performed and interpreted with similar accuracy. Intravenous contrast may lead to nephrotoxicity in at-risk individuals, is unnecessary to accomplish certain diagnostic tasks and may be reduced for others [35]. Lower tube potential and noise reduction techniques are newer methods to permit reductions in intravenous contrast and radiation dose without compromising diagnostic benefit in many situations (see “Optimization, individualization, and implementation”) [35–40].

Communication with referring clinicians and patients

Clearly, communication between patients, clinicians, and radiologists is critical to ensure a CT exam is medically justified and appropriate for each patient’s individual context, not to mention that s/he should understand the procedure and potential benefits of the exam. Patients are aware of idiosyncratic factors, such as their symptoms and family history and their willingness to undergo an exam (oral contrast, bowel preparation, IV stick, claustrophobia, fear of radiation) that may affect which types of exams are likely to provide the most benefit. Referring clinicians are likely best positioned to estimate the net benefit from CT examination as they can weigh the likelihood of potential disease(s) for a particular patient, its potential severity, the ability of CT to detect and characterize or stage the disease, the impact of alternative therapies, potential impacts of false negative and false positive exams, and the risk of not obtaining a CT exam. Radiologists can answer questions about test performance and radiation risk, and certainly need to understand the concerns of the referring clinician when optimizing the CT acquisition methods and parameters, and the potential need for post-processing and/or subspecialized interpretation. Unfortunately, a recent survey of a multispecialty practice found that only 30% of clinicians discuss the risks and benefits of CT imaging with their patients prior to obtaining a scan, potentially in part because they do not understand the risk of radiation themselves [32]. Discussion of CT benefit and risk can be facilitated proactively by radiologists by providing resources to referring clinicians and answering questions as soon as they arise. Radiologists can provide referring clinicians with pertinent and individually relevant resources that focus on common types of CT examinations, so that referring clinicians are able to have relevant knowledge at hand when considering, ordering, and discussing CT exams with their patients. Some radiology practices discuss exam appropriateness with different subspecialties and provide pamphlets and materials to referring clinicians to facilitate this discussion [41]. An example of bullet points and individually pertinent facts facilitating clinician–patient discussion of CT benefit and risk is provided for CT enterography and colonography (Figs. 3, 4). Other resources may take the form of decision support systems, hyperlinks to ACR appropriateness criteria [42], or institutionally agreed upon guidelines for common imaging questions (for example, imaging of the pregnant patient).

Fig. 3
figure 3

Individually relevant facts facilitating clinician–patient discussion of CT benefit and risk is provided for CT colonography.

Full size image
Fig. 4
figure 4

Individually relevant facts facilitating clinician–patient discussion of CT benefit and risk for CT enterography in patients with known or suspected Crohn’s disease.

Full size image

When one considers types of CT examinations that were unavailable or not routine two decades ago, it becomes obvious that perceived patient benefit largely drives CT justification: for example, dual-energy for uric acid stone detection and treatment; CT enterography for detection of Crohn’s disease and its complications; CT urography for known or suspected transitional cell carcinoma; CTC to detect colorectal polyps and cancers; multiphase CT enterography to detect small bowel cancers and masses in occult GI bleeding are all relatively new diagnostic tasks for CT imaging that are linked with potential efficacious therapies [4, 6, 7, 43, 44].

Conversations between clinicians and radiologists can lead to some important but uncomfortable conversations. For example, it may be necessary to explain that larger cumulative doses are justified in very sick patients (for example, severe acute pancreatitis—where the risk of organ failure and sepsis is high) or when the risk of misdiagnosis is high (for example, detection of pancreatic cancer or HCC—where potentially increased radiation risk is offset by the higher risk of death). In addition, dose reduction hardware and software can be very expensive. Most departments will be unable to deploy these technologies throughout their entire practice, so it will be necessary to strategically place these technologies where they can provide the most benefit, and/or triage particularly important groups of patients (such as children) to this technology. It will likely require some time for radiology practices to sort out these types of tradeoffs for themselves, much less discuss them with referring clinicians. In addition, while patients with many CT exams may not wish to receive additional CT imaging, a necessary correlative to the linear no-threshold hypothesis is that any cumulative threshold is completely arbitrary and without clinical meaning—the risk for the 10th CT exam should be the same as the first. Justification should be based on clinical history and anticipated benefit of considered exam, regardless of prior exposures [45].

Optimization, individualization, and implementation

Exam optimization is the responsibility of the entire imaging team, composed of radiologists, technologists, and physicists. It includes thoughtfully selecting acquisition methods and interpretive tools for a CT exam so that disease detection is maximized while minimizing dose and non-radiation risk [46]. However, because the magnitude of net benefit is driven by the potential benefit of a positive exam, radiation dose should not be reduced if doing so may compromise accomplishment of the diagnostic task. For example, increasing CT slice thickness and decreasing dose is not beneficial to the patient if it means that a pancreatic cancer could be missed (Fig. 5). Other CT acquisition parameters, such as the phase of enhancement, tube potential selection, and oral and intravenous contrast can dramatically impact the ability to identify disease for any given diagnostic task [39, 47]. Imaging parameters must also be tailored to individual patient size using tube current modulation, tube potential selection or technique charts, and consideration of the patient’s age [48]. The easiest and most important way to reduce radiation dose in large and small CT practices is to eliminate CT exams that are not clinically indicated, as well as superfluous acquisitions (such as routine unenhanced or delayed scans), that do not contribute to making an accurate medical diagnosis.

Fig. 5
figure 5

Routine portal-phase contrast-enhanced CT with 6-mm slice thickness at level of pancreatic head and body (A, B). Exam interpreted as negative by non-abdominal radiologist. Due to continuing clinical suspicion and re-evaluation by GI radiologist, exam was repeated 2 days later with dual-phase technique, increased tube current, and 2.5-mm slice thickness, demonstrating improved visualization of low attenuation mass in pancreatic head (C) and dilated pancreatic duct (D) and leading to diagnosis of pancreatic cancer.

Full size image

Several challenges exist to optimization. There is lack of information on the specific image quality requirements (e.g., spatial and contrast resolution) needed to achieve acceptable performance for specific diagnostic tasks, so radiologists are often left to utilizing the dose levels used in large retrospective or prospective studies. Generally, radiation dose can be greatly reduced when the image contrast differences needed for diagnosis are very high (renal stone identification, CTC, CT angiography), but radiation doses will need to be higher when subtle soft tissue attenuation differences need to be detected (e.g., liver mass identification/characterization). In addition, a frequently overlooked aspect to optimization is the expertise of the interpreting radiologist (Fig. 5). This consideration is essential in terms of obtaining potential benefit from CT exams. A GI radiologist may be attuned to the imaging findings of autoimmune pancreatitis that can be seen on even a four slice scanner, but not attuned to the imaging findings of medullary sponge kidney of CT urography. Similarly, software infrastructure must be established throughout departments so that appropriate image reconstruction and post-processing facilitate the accomplishment of diagnostic tasks, such as dual-energy post-processing or interactive 3D imaging for CTC. In the current age of CT dose reduction with noise reduction techniques, departments will need to carefully evaluate and compare image-based vs. scanner-based noise reduction approach. Scanner-based alternatives may have less impact on workflow, but may be more expensive to implement in a large department.

In this age of dose awareness, departmental issues to be addressed by radiology departments include the triage of appropriate patients to low dose techniques, and implementation of low dose protocols throughout the CT practice. Observer performance studies demonstrate that radiologists are very good at evaluating noisy low dose images without an impact on disease identification or staging for a variety of diseases, so expensive technology is not always needed [49–53]. Radiologists across a department need to agree upon standard protocols for common clinical situations. Thus, the transition to a lower dose CT practice involves efficiency, reproducibility, and communication. It also involves weighing potential alternatives, such as an investment in decision support systems (to eliminate unnecessary exams) vs. investing in noise reduction hardware and software.

Conclusion

Dose awareness and communication by radiologists and radiology departments can be beneficial for patients, clinicians, and radiology departments. Initiating and facilitating discussions of the net benefit of CT by enlisting comparisons to more familiar activities (swimming or bicycling), or by conveying that the anticipated radiation dose to an exam is similar to or much less than annual background levels helps resolve the concerns of many patients and providers. For example, our pediatric oncologists had eliminated surveillance chest CT and performed surveillance chest X-ray instead, until we made a low dose chest CT exam available to them. Demonstrating that these CT exams could be performed using radiation dose levels similar to 3 weeks of natural background exposure in Rochester, Minnesota, overcame their reservations. Similarly, implementation of low dose CT enterography across our practice has helped our gastroenterologists to understand that we are considering patient risk and benefit in conjunction with them when scanning their Crohn’s disease patients.

While radiation risk estimates regarding CT contain considerable uncertainty, we choose to err on the side of safety by assuming a small risk exists in order to protect our patients. We also understand that the benefits and risks of CT are highly individualized and require consideration of many factors by patients, clinicians, and radiologists; however, many of these factors can be understood and communicated before CT exams are ordered, so that appropriate discussions can occur with the patient. Radiologists should assist clinicians and patients with understanding test performance, potential benefit, and estimates of potential risk. Radiologists should individualize CT scans according to patient size and diagnostic task, to ensure that maximum benefit and minimum risk can be achieved, and should work across departments so that a standard and thoughtful approach can be taken for every patient.