Since its inception in 1973, the role of CT in diagnostic radiology has expanded. According to surveys conducted at United States medical facilities, the annual number of CT examinations increased from approximately 3.6 million in 1980, to 13.3 million in 1990, and to 33 million in 1998 [1, 2]. Currently, about 13% of all radiology procedures in the United States are CT (Sunshine J, personal communication). Not only has the use of CT increased markedly in the past 20 years, but its technical capabilities also have improved dramatically [3]. The development of slip rings, increased X-ray tube heat capacity, advances in detector technology, and improvements in computers have permitted rapid subsecond axial and helical CT scans. Single- and multislice helical CT can image large anatomic regions in a single breath-hold. CT fluoroscopy has provided a real-time tool for imaging-guided biopsy procedures. Improvements in software and display systems have led to virtual CT colonoscopy, CT angiography, and a number of other clinically useful applications.

Inherent in the design of advanced CT scanners providing many new applications are elements that have the potential to increase radiation exposures to patients. This potential certainly applies to the new multislice CT scanners appearing in the work-place. Various publications have estimated the typical surface radiation doses to adults from multiple adjacent CT slices as 30-70 mGy (3.0-7.0 rad) per head scan series and 20-50 mGy (2.0-5.0 rad) for each abdominal series [4,5,6,7,8]. For standard measurements with phantoms, the head radiation dose is nearly uniform; the body radiation dose is essentially uniform over the surface and decreases to about half at the center [4,5,6,7,8]. If adult scan parameters are used to scan pediatric patients, typical radiation doses are at least double the adult values [9, 10]. CT studies that overlap the scanned regions or rescan the same anatomy (such as enhanced and unenhanced scans) can have 2-3 times the radiation doses of nonoverlapped scans. CT fluoroscopy typically delivers dose rates around 20-60 cGy/min (20-60 rad/min) [11,12,13]. To place these values in perspective, the patient surface dose from a typical abdominal CT is 200-300 times more than that of a typical chest radiograph, 20-30 times that of the mean glandular dose of a typical craniocaudal mammogram, and approximately 10-20 times that of typical abdominal radiography [14]. In 1980, when CT accounted for only 1.8-2.5% of all radiologic examinations, CT was calculated by the National Council on Radiation Protection to deliver about 5% of the collective dose from all X-ray procedures [15]. On the basis of that calculation and of the fact that CT now accounts for 13% of all examinations, we estimate that CT currently contributes approximately 30% of the collective dose in the United States. For comparison, in the United Kingdom, CT accounts for only 4% of all radiologic examinations and delivers more than 40% of the total radiation dose [16, 17].

Because of the high radiation exposure potential of present-day helical and multislice helical CT, radiologists should be aware of the radiation risks of CT and work actively to keep patient radiation exposures from CT as low as possible while achieving the required image quality and medical benefit. Radiologists also need to promote patient and public awareness of the radiation risks related to CT, so that the public realizes that radiologists are responsible users of the modality and guardians of the public health. Media hyperbole about radiation risks may make a positive public perception of CT difficult to achieve and sustain. Such media hyperbole easily elicits fear from the public, as it did recently when the issue of potential overexposure of children by CT was overzealously reported [18]. To be ready for the concerns that will inevitably continue to be voiced by patients, radiologists should become more aware of the issues and controversies surrounding the public concern that CT will cause cancer.

Controversy exists about the carcinogenic potential of the relatively low levels of ionizing radiation exposure associated with CT. The relationship of these radiation exposures to biologic risk for patients is determined by mathematic extrapolation based on changes observed after exposure to much higher levels of radiation. The potential risk is calculated with linear or linear-quadratic extrapolations from the higher doses [19]. Experts argue over whether this estimation is too conservative, but in the best interest of protecting patients, such conservative estimates probably can be justified. Using such approaches, the Committee on Biological Effects of Ionizing Radiation estimates a 1.2-1.5% increase in fatal cancers for 5-year-old children who receive a uniform body dose of 100 mGy (10 rad) [19, 20]. The increase in breast cancer for 15-year-old girls who receive a breast dose of the same magnitude is about 0.3%. Similarly, it is estimated that the delivery of 1 rad (0.01 Gy) to the breast of a woman less than 35 years old increases the risk of breast cancer by about 14% over the spontaneous rate for the general population [21]. There also are special risks for fetal exposure with theoretic increases in rates of leukemia and other forms of malignancy. Thus, given application of the linear hypothesis for risk assessment, even though the absolute radiation exposure from CT is small, we cannot consider the risks negligible.

How should a radiologist apply this information about CT so that it benefits patients? First, scanning techniques can be altered to provide acceptable images at lower radiation exposures. Possible adjustments in the tube current, the spacing of CT slices, and slice thickness could reduce the radiation exposure while maintaining diagnostic image quality. The radiation dose to the patients from CT is directly proportional to the tube current and scan time (however, scan time is usually short and constant to minimize motion blur). For pediatric patients, the product of amperage and scan time should be significantly less than that routinely used for adult studies [22]. Because of the smaller pediatric anatomy, image quality is not likely to be compromised by this adjustment in scan parameters. The same logic applies for CT fluoroscopy; the amperage and total scan time usually can be reduced for both pediatric and adult patients without substantial degradation of studies. Some CT units have selectable scan modes in which the X-ray tube current is continuously modulated online on the basis of measurements of the patient's tissue attenuation [23]. Use of this feature would adjust the necessary radiation levels to compensate for patient size and the anatomy being imaged; considerable savings in the radiation dose can be achieved with this approach.

The spacing of CT slices also is important and is described by the pitch. The pitch is the table incrementation per one X-ray tube rotation divided by the CT slice collimation. For a single-slice axial or helical CT, a pitch of 1:0 means that the slices are adjacent and not overlapped. A pitch greater than 1:0 means there are gaps between the slices; the gaps reduce the radiation dose by averaging less exposed tissue in the gaps between CT slices with the irradiated tissue. A pitch of less than 1:0 results in an overlap of the scanned tissue that increases the radiation dose.

Using a large number of thin adjacent CT slices results in 30-50% more radiation dose to the patient than using fewer thicker slices to scan the same anatomy. In CT scanners, radiation extends beyond the slice collimators because of scattered radiation, focal spot penumbra, and the cone-beam geometry of the X rays. For many thin CT slices, this radiation extends outside the imaged slice and overlaps to a greater degree than for thicker CT slices. Moreover, because the image noise is greater with thin slices, frequently the amperage is increased to compensate. This increase raises the radiation exposure further.

For multislice CT scanners, the pitch values must be interpreted differently. For example, the table may be incremented 30.0 mm to image several simultaneous 10.0-mm slices; however, if the CT unit collimates the X-ray beam for four 10-mm slices simultaneously (effective pitch, 0.75), there is an overlap of the irradiated tissue. Regardless, multislice CT scanners usually have several modes in which the user can elect to either overlap or leave gaps between slices. Generally, the radiation doses to patients are about 30-50% greater with multislice CT as a result primarily of scan overlap, positioning of the X-ray tube closer to the patient, and possibly increased scattered radiation with wider X-ray beams [24, 25]. Patient radiation dose can be reduced by decreasing the mAs and increasing the pitch to greater than 1:0. It has been suggested that pitch values up to 1:5 do not degrade image quality [26, 27]. Moreover, because of the increased radiation dose, users should be discouraged from using the speed capabilities of multislice CT to perform redundant thin-section studies of the same anatomy.

Increasing the X-ray tube kilovoltage increases both the radiation dose and the penetration of the X rays through the body. With all other scan parameters kept constant, an increase from 120 to 140 kVp generally increases the patient radiation dose by 30-40%. Accordingly, increases in kilovoltage should be avoided. However, an increase in X-ray tube kilovoltage could be accompanied by a reduction in tube current to diminish this effect.

In addition to these adjustments of technical parameters, radiologists should consider developing more well-defined approaches to selection of patients for CT, which should be used for the myriad indications for which it is expected to provide medical benefits. In situations in which the diagnostic yield of CT is expected to be low, however, alternate examinations should be considered. On the basis of known biologic risks, we suggest that three types of CT should be of particular concern: CT of children, thoracic CT in which the field of view includes the female breasts, and abdominal CT of pregnant patients [28].

The pediatric issues have been covered effectively by others [21, 23, 29]. Sonography and MR imaging should be considered alternate abdominal imaging approaches in pregnant patients. The issue of breast exposure during thoracic CT is relevant for many potential thoracic applications of CT. Exposures of 2.0-5.0 cGy (2.0-5.0 rad) at the body surface are typical for conventional thoracic CT [14]. This dose compares with an average glandular breast dose of 3 mGy (300 mrad) for a typical two-view screening mammography, and a body surface dose of 2-4 mGy (200-400 mrad) for low-dose chest CT for cancer screening. [14]. Multislice CT of the heart for calcium scoring can deliver substantially higher radiation doses (up to five times the typical thoracic CT levels) because multiple exposures are required to gather data during different periods of the cardiac cycle. Dose rates for typical CT angiography are in the same range as the typical thoracic CT (2.0-4.0 cGy). If electron-beam CT is used for CT angiography or for cardiac calcium scoring, the anterior chest wall and breast dose is substantially lower because the X-ray beam originates behind the patient. Remy-Jardin and Remy [21] and Mayo et al. [30, 31] have recommended decreased kilovoltage settings, wider spacing of slices such as a pitch of 2, and possible use of bismuth radioprotective garments [32] to reduce the breast dose during CT angiography with helical scanners.

In low-yield settings, especially when dose reduction strategies are not in use, consideration of alternate diagnostic imaging approaches to CT angiography may be pertinent. For example, a middle-aged woman who presents with chest pain associated with pulmonary embolism, but is thought to have a low prior probability of pulmonary embolism, could be selected instead for scintigraphy (which gives a radiation exposure <1 cGy to the breasts) if findings on her chest radiograph are normal or nearly normal. Likewise, should CT be used to screen otherwise asymptomatic healthy adults in attempts to find early cancer? Such approaches now are widely marketed in certain parts of the country but are not accompanied by data to show significant yields of detecting early cancer. The likelihood of false-positive findings that might lead to further CT or other procedures enhances the possibility for repeated exposures. At least asymptomatic patients who undergo such procedures should receive written and verbal information about the estimated increase in cancer risks associated with the procedure that they are undergoing to reduce their cancer risk. Such information and documentation of the patient's full knowledge also should help avoid unnecessary legal confrontations in the future should such patients develop cancer. Given that the latency time for cancer induction in the dose ranges used in CT is calculated to be 10-30 years, policies need to be adopted now to avoid problems in the future.

Regarding other policies that might help control radiation exposure from CT to patients, guidelines being created by organizations such as the American College of Radiology, the Society of Pediatric Radiology, and the Fleischner Society may help. The American College of Radiology has been developing a CT accreditation program for approximately 2 years. This program contains various elements related to equipment and radiation safety, including a module on pediatric CT. The American College of Radiology estimates that this program will be available soon (Amis ES, personal communication). In addition, we recommend that a qualified radiology physicist determine and post the estimated patient radiation doses for various CT scans in all CT working areas. This notification will serve to remind technologists and radiologists about judicious selection of scan parameters.

Radiologists should take a strong position regarding the meticulous control of radiation exposures to patients from diagnostic CT. Actions that we advocate to achieve this goal include making available to patients information brochures about the benefits of CT, the relatively low risks of radiation exposures from CT, and how those risks are being reduced by safety-conscious approaches to the use of CT. Radiology educators should make instruction in these areas a high priority. The American Board of Radiology should include more testing on this subject in examinations. Industry should be requested to provide detailed dosimetry calculations of all the various CT modes in their equipment as part of the written documentation of their equipment. Such actions will allow radiologists to maintain the public's trust and to be viewed by the public as responsible providers of CT services.