The contribution of computed tomography (CT) has been remarkable in its scope. Its introduction in the 1970s enabled volumetric scanning in the form of a series of single-slice incremental acquisitions along the longitudinal axis (the z-axis) of the patient. A bank of contiguous detector elements was arrayed along the arc of the gantry, enabling interrogation of the tissues in the fan-shaped beam path at multiple angles. The signals generated in the detectors by x-ray quanta—differentially attenuated by the constituent tissue types irradiated—become the raw data for image reconstruction.

A significant limitation of single-slice incremental scanning is the partial-volume effect. In this approach, the data might be incompletely volumetric, in that gaps could exist between the irradiated tissue slabs. This results in the potential for failure to capture small lesions, which could be situated only partially within the interrogated volume, with reduced contrast, or in-between consecutive slices. Further, conventional single-slice scanning required several minutes to complete, and scan protocols were limited in both tube current and scan length by the x-ray tube’s heat capacity.

The introduction of helical CT in the mid-1990s enabled true volumetric single-slice scanning with no interslice gap. In this mode, data is acquired continuously and in synchrony with patient translation using high-power x-ray tubes and interpolative reconstruction algorithms to correct for noncoplanar (helical) projection data. As complete scans could be accomplished in a single breath hold, clinical applications of CT as a diagnostic modality grew to encompass examinations of anatomy requiring rapid interrogation to avoid either blurring from organ or patient movement.

Multidetector CT (MDCT) systems, introduced within the past decade, currently feature 2 to 64 multiple detectors arrayed along the z-axis. This innovation enables simultaneous acquisition of attenuation data corresponding to multiple contiguous slices with a concomitant reduction in scan time. Associated with the emergence of MDCT has been the development of submillimeter slice widths, rendering high spatial resolution in all planes—in other words, isotropic resolution. Submillimeter imaging has enabled expansion of MDCT applications to include those typically addressed by planar imaging modalities, such pulmonary angiography for pulmonary embolism, CT pneumocolonography, and oncologic staging. Further, this capability is a prerequisite for 3-D and multiplanar reconstructions of sufficiently high image quality for such applications as CT angiography.

Special Considerations

With the clinical utility of these innovations firmly established, the frequency of CT examinations has increased more than fivefold worldwide in the past decade. Over that same time interval, CT has made a steadily incremental contribution to the collective dose of the population. Although CT examinations constitute approximately 11% of all x-ray procedures in the United States, the modality has been estimated to deliver more than two thirds of the total radiation dose from all sources of radiological imaging using ionizing radiation.

The health detriment resulting from exposure to ionizing radiation generally are classified as deterministic or stochastic. Deterministic injuries result from cell death in significant numbers—such as is involved in the production of radiogenic skin injury, such as erythema—and have characteristic dose thresholds, typically on the order of several hundred rads. (Although such injuries are rare in diagnostic procedures, they are most commonly seen in fluoroscopically guided interventions. With the advent of CT fluoroscopy, however, the potential for delivery of significant skin doses exists.) Stochastic effects include the production of cancer as well as heritable genetic mutations. The risk of development of a radiogenic health detriment is typically expressed in terms of effective dose. For x-ray procedures, effective dose—measured in units of Sieverts (Sv)—is the absorbed dose to the whole body weighted according to the radiation sensitivity of the different tissues irradiated. Absorbed dose (energy absorbed per unit mass) to any given tissue volume is typically reported in units of Gy, where:

1 Gy = 100 cGy = 100 rads

A number of different dose metrics specific to CT exist, the principal one being the computed tomography dose index (CTDI). Developed for application in single-slice axial CT, the CTDI quantifies the average dose to a single slice in a series of contiguous scans. It does not account for either scan gaps or overlaps, as could occur in helical MDCT scanning. A different metric, CTDIvol, captures the average dose delivered to the scanned volume, and is calculated as CTDIvol = CTDIw/pitch (where CTDIw is the average CTDI in the scan plane, that is, the differently weighted average of central and surface doses for a given protocol). Effective dose is related to these volumetric dose indices through experimental determination of conversion factors associated with scanning protocols for different body sections. These conversion factors are typically reported in units of effective dose:

(mSv) / dose-length product (DLP)

where

DLP = CTDIvol x scan length

Special consideration regarding dose at CT must be given to pediatric applications. Children in particular incur elevated risk relative to adults owing to greater radiosensitivity and longer life span over which radiogenic injury could manifest. Further, due to smaller organ size in general, relatively greater quantities of energy are imparted per unit mass (that is, greater dose) for a given technique. This same consideration holds true for small adults.

A number of factors contribute to the dose intensity of CT:

1) Mode of irradiation. In contrast to planar radiography, which is planar by nature, CT involves multi-angular irradiation of the patient. As a result, dose is distributed with essentially uniform intensity throughout the scan plane rather than with the decreasing intensity with depth characteristic of radiography. In general, a CT examination of a given section of anatomy delivers a dose that is substantially higher than its radiographic equivalent. For example, the effective doses delivered by radiography and CT of the (PA) chest are approximately 2 and 800 mrem, respectively. Order-of-magnitude dose disparities between CT and radiography exist for other examination types.

2) Dose to extraneous tissues. Relatively high doses are delivered to tissues included in the scan plane but not of clinical interest, such as the breast in thoracic CT. Breast dose during such procedures lies in the range of 2 to 10 rad—in comparison with an average mean glandular dose of approximately 200 millirad per view in mammography. Dose to all tissues in the field of view at CT as well as those irradiated by secondary radiation (internal and external scatter and tube leakage) contribute to the patient’s effective dose.

3) Irradiated tissue volume. With the advent of helical MDCT and subsecond gantry rotation times, and the option of contiguous or overlapping scans, greater scan lengths are achievable in increasingly less time—resulting in a concomitant increase in the average total volume of irradiated tissue. Further, the requirement for interpolation of transmission profiles from neighboring scans in helical scanning in turn necessitates additional rotations of the gantry at the extremes of the scan range, such that the exposed tissue volume is greater than the reconstructed volume. This type of dose augmentation is exacerbated as the aperture width increases.

4) Nature of CT image formation. Modalities that use image-recording media with limited dynamic range—such as screen-film radiography—have associated with them limits to the dose that can be recorded without loss of information. CT, however, is an inherently digital-imaging modality for which there is no such dose penalty. Image quality in CT will increase with increasing dose as the level of Poisson-distributed noise decreases.

5) Nonoptimized scanning protocols. The NRBP UK CT dose survey1 demonstrated that patient-efficient doses for the same examination could vary by up to a factor of 10 among institutions. However, this magnitude of variability represents a significant improvement in the findings of the 1991 survey by the same group in which dosewise variation on the order of a factor of 40 was found. This change is, in part, the result of emergent awareness of the radiation burden imposed by CT and the application of dose-mitigating strategies based on patient age, body habitus, and the tissue type to be imaged.

Dose Management

Given the proven utility of CT in providing diagnostic information for a panoply of malignant and benign disease, the goal of dose management in the context of medical imaging involves dose optimization rather than simply reduction—the end point being providing maximum diagnostic information content (that is, image quality) at the lowest possible dose. The relationship between image quality and dose in CT is relatively complex, involving the interplay of a number of factors, including noise, axial and longitudinal resolution, and slice width. What follows is a basic description of those parameters that impact dose and image quality, as well as provide a template for objective evaluation of different scanners or protocols in the framework of dose optimization.

1) Scanner-Design Parameters.

Scanner geometry. Scanner geometry—specifically, the focus-to-isocenter and focus-to-detector distances—determines the photon fluence incident upon the patient and, therefore, is useful in the creation of an image. A “short geometry” scanner will, by virtue of inverse-square-law effects, provide a larger number of x-ray photons and a higher-quality image—at a higher dose—than an otherwise equivalent “long-geometry” scanner using the same scan prescription. Differences in scanner geometry, then, must be taken into account with the application of a given scan protocol to different scanner models.

Detector geometric efficiency. All multidetector scanners have in common simultaneous measurement of multiple dose profiles arrayed contiguously along the z-axis of the patient. Due to difficulties encountered in calibration of multi-detector systems resulting from exposure of the outermost detectors to the focal spot penumbral (shadow) radiation, these systems typically employ “overbeaming,” wherein the irradiated-section width is wider than the active detector width. In these systems, then, geometric efficiency—the ability to capture all available incident information (x-ray quanta) for image formation—is less than 100%. For geometric efficiencies less than 100%, the penumbral dose is “wasted” with respect to image reconstruction, but contributes nonetheless to patient dose.

The relative contributions of umbra and penumbra are determined, in part, by scanner geometry—specifically, the distances from the focal spot and beam-shaping filter. Ultimately, the penumbra extends 1.0 to 1.5 mm on either side of the full beam (umbra); the geometric efficiency of single-slice scanners is, in general, 100%. For multidetector scanners, the geometric efficiency is a strong function of aperture width (nominal section thickness x number of active detectors), with narrow widths being the least efficient.

Geometric efficiency can be approximated for any scanner type and detector configuration as:

aperture width (mm)/[aperture width + 2x penumbra width]

Assuming a penumbral width of 1.5 mm, a section thickness of 0.5 mm, and collimation (aperture) width of 2 mm (that is, a detector configuration of 4 x 0.5 mm—typical of a four-slice scanner), the geometric efficiency would be approximately 40%. Note that for a 16-slice scanner using an aperture width of 8 mm for the same section thickness (0.5 mm), the geometric efficiency would be greater than 70%.

Figure 1. Decreasing kVp can produce more rapid deterioration in image quality than do decreases in mAs of comparable magnitude.

2) Scanning Parameters.

Tube potential and current. Radiation output and beam penetrability are proportional to scan kVp. Specifically, output increases as kVpn such that significant modifications in dose from CT exams can be effected by reductions in kVp. For example, a reduction of kVp from 120 to 80 in pediatric abdominal scanning would produce a dose decrement—all other parameters being equal—of more than a factor of 3. However, reduction in kVp could result in an unacceptable increase in noise without a compensatory increase in mA, and could be entirely unacceptable when imaging large habitus patients and/or thick sections of anatomy, or when imaging tissues or disease with inherently very low contrast. If mA is increased, it should be increased only enough to achieve the same level lesion detectability while achieving a lower dose. However, in Figure 1, it can be seen that decreasing kVp can produce more rapid deterioration in image quality (here, measured as the contrast-to-noise ratio) than do decreases in mAs of comparable magnitude.

Rotation time. The combination of subsecond scanning speeds and submillimeter scan widths, made possible by advances in MDCT gantry technology, presents a unique problem in dose optimization. Radiation exposure of any scanned section is proportional to the product of mA and gantry rotation time (mAs). Although rapid scan speeds reduce the overall exposure time and dose, the noise level will increase without an increase in tube current. Increased noise levels will reduce the conspicuity of small lesions. However, as indicated previously, thin-section scanning allows for alleviation of the partial-volume effect, which could serve to offset the increase in noise.

Relationship between dose, aperture width (collimation), table speed, pitch, and interpolation algorithm. For helical MDCT, the concept of effective mAs (mAs/pitch) is useful for elucidating the relationship between scan technique and radiation exposure. Here, pitch is defined as the ratio of table feed per gantry rotation to aperture width. Pitch greater than 1 decreases the exposure to a given scanned section, but with a loss of resolution and increase in noise. Pitch less than 1 increases exposure due to overlaps in trace of the helical beam. Effective mAs, then, can be used to construct a scale of relative dose, as patient dose increases with the total volume irradiated, which is, in turn, proportional to effective mAs. For MDCT, the number of data sets along the z-axis is variable for different pitches. Data is interpolated over a fixed distance, rather than a fixed number of data points (a process referred to as z-filtering), such that the same slice width and noise characteristics can be achieved regardless of pitch.

Reconstruction filters. Image noise and spatial resolution are strongly determined by choice of convolution kernel. These filters, to varying degrees, effect noise suppression and resolution enhancement in a manner specific to the imaging task—standard, bone, soft tissue, and lung, for example. The ability of a particular reconstruction filter to reproduce the inherent subject in the image as a function of spatial frequency is quantified by the modulation transfer function (MTF).

Approaches to Dose Optimization

Quantitative. Various figures of merit (FOM) have been developed to quantify image quality based on resolution, dose, section width, and noise. One such FOM is Q, which numerically relates these parameters to image quality as

Q = √ [f3 av2 z1 CTDIvol]

where f3av is the average spatial resolution (that is, the average of 10% and 50% MTF values), s is the percent image noise, and z1 is the width of the imaged slice profile. This approach requires measurement of noise and spatial resolution for a range of reconstruction filters at a fixed dose (CTDIvol) and slice width. One approach to the goal of dose optimization is addressed when the dose at which noise and resolution are measured is constrained to be less than or equal to American College of Radiology reference dose levels (head [cerebellum]: 60 mGy; adult body: 35 mGy; and pediatric body: 25 mGy).

Factors Contributing to CT Dose Intensity

  1. Multi-angular mode of irradiation.
  2. Dose to extraneous tissue.
  3. Irradiated tissue volume increases with width of aperture and contiguous scans.
  4. Digital imaging does not carry a dose penalty: image quality increase with dose level.
  5. Imaging providers are not optimizing scanning protocols.

Applied. In the US Food and Drug Administration Public Health Notification of November 2001,2 it is stressed that significant CT dose mitigation can be accomplished by facilities by application of fundamental dose management concepts:

  1. optimize scan parameters by reducing tube current, either as a user-selectable option in the scan prescription or by selecting scanners that feature angular and z-axis modulation of mA according to tissue thickness along a given projection;
  2. develop tabulated guidelines for height- and weight-adjusted current settings;
  3. increase table increment or pitch (axial and helical scans, respectively) whenever possible without reducing image quality;
  4. reduce the number of multiphase scans; and
  5. reduce inappropriate referrals and recommend less dose-intensive modalities—such as MRI, ultrasound, and radiography—whenever possible.

Ultimately, CT dose management requires a team of professionals—technologists, physicians, administrators, and medical physicists—to ensure that the most effective and judicious use of this remarkable diagnostic tool.

References

  1. Shrimpton PC, et al. Doses from computed tomography (CT). Examinations in the UK-2003 Review. NRPB-W67. National Radiation Protection Board, London, UK, 2005.
  2. Public Health Notification: Reducing radiation risk from computed tomography for pediatric and small adult patients. US Food and Drug Administration. November 2, 2001. Available at: FDA Public Health Notification. Accessed May 9, 2006.

Additional Sources

  • ACR CT Accreditation Program Requirements. American College of Radiology, Reston, Va. 2003.
  • Crawley MT, Booth A, Wainwright A. 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. Br J Radiology. 2001;74(883):607?614.
  • Edyvean S, et al. Presentation: A methodical approach for comparison of CT image quality relative to dose. ImPact (Imaging Performance Assessment of CT Scanners). St George’s Hospital, London, UK. Presented at RSNA, Chicago, Ill, November 30, 2003.
  • European Commission consortium report: Framwork research project of the European Commission: CT Safety and Efficacy. 2004.
  • Kalra MK, Maher MM, Toth TL, et al. Strategies for CT radiation dose optimization. Radiology. 2004;230:619.
  • Lewis MA, Edyvean S. Patient dose reduction in CT. Br J Radiology. 2005;78:880?883.
  • Mahesh M. Search for isotropic resolution in CT from conventional through multiple-row detector. Radiographics. 2002;22:949?962.
  • mPACT, Four Slice CT scanner comparison report, version 11, Medicine & Healthcare products Regulatory Agency MHRA 03023, London, UK, 2004.

Cynthia Goodman-Mumma, MS, MSE, DABR, is diagnostic imaging physicist and radiation safety officer at Lehigh Valley Hospital Health Network (Allentown, Pa).