Published online before print November 7, 2006, 10.1148/radiol.2421051350
(Radiology 2007;242:208-216.)
© RSNA, 2006
Overranging in Multisection CT: Quantification and Relative Contribution to Dose—Comparison of Four 16-Section CT Scanners1
Aart J. van der Molen, MD and
Jacob Geleijns, PhD
1 From the Department of Radiology C-2S, Leiden University Medical Center, Albinusdreef 2, NL-2333 ZA Leiden, the Netherlands. Received August 12, 2005; revision requested October 18; revision received January 4, 2006; accepted February 1; final version accepted April 25.
Address correspondence to A.J.v.d.M. (e-mail: molen{at}lumc.nl).
 |
ABSTRACT
|
|---|
Purpose: To quantify the number of overrange rotations and to assess their relative contribution to organ and effective doses at 16-section body computed tomography (CT).
Materials and Methods: Overranging was quantified for four 16-section scanners by means of free-in-air dose measurements at different scan lengths. Overrange rotations and lengths at a certain section width were derived for all collimations and clinically used pitches by extrapolation. The effect of reconstructed section width on overranging was analyzed separately. Results were applied to clinical protocols for the chest and abdomen. Thyroid and testicular dose and effective dose were established, and relative dose contributions from overranging were calculated. Statistical analysis was performed by using Pearson correlation and paired t tests. P < .05 indicated a significant difference.
Results: The number of overrange rotations showed considerable differences between scanners, with a range of 1.99–4.04 at the lowest and 0.93–2.59 at the highest pitch. Number of rotations correlated negatively with pitch, while overrange length correlated positively with collimation and pitch. The effect of section width was variable. In the protocols, overrange length ranged from 3.2 to 5.8 cm for chest and from 3.2 to 5.2 cm for abdominal CT. When the contribution of overranging was not taken into account, significantly lower values for thyroid (P = .012) and testicular (P = .025) doses and effective doses for chest (P = .005) and abdominal (P = .011) CT resulted.
Conclusion: Overranging is reconstruction-algorithm specific, and its length generally increases with collimation and pitch, while the effect of section width is variable. Overranging may lead to substantial but unnoticed exposure to radiosensitive organs.
© RSNA, 2006
|
INTRODUCTION
|
|---|
Overranging, or z overscanning, is an integral part of single- and multisection computed tomography (CT). Because in helical scan mode the reconstruction algorithm requires additional raw data on both sides of the planned scan, extra rotations outside the planned length are needed for image reconstruction (1). To enable reconstructions with both thin and thick section widths from the same raw data, the choice of reconstruction parameters also plays a role. Both factors will lead to exposure of tissue volumes above and below the planned length. This adds to the radiation dose to the patient and is therefore important for accurate dose calculations. Proper CT protocol optimization with careful selection of scanning parameters can minimize overranging and the dose contributions associated with it.
In single-section spiral CT, the number of overrange rotations is well known—for example, 1.0 and 2.0 for the 180° and 360° linear interpolation algorithms, respectively (2).
In multisection CT, image reconstruction algorithms are more variable and scanner specific, and few data on overranging exist (1,3). With wider beam collimations in the newest scanners, the additional length from overranging and its relative contribution to radiation dose may increase accordingly if its contribution outweighs the improved geometric dose efficiency of these scanners. Understanding this phenomenon is becoming increasingly important. Thus, the purpose of our study was to quantify the number of overrange rotations and to assess their relative contribution to organ and effective doses in 16-section body CT.
|
MATERIALS AND METHODS
|
|---|
Other than providing protocol suggestions, the CT scanner vendors offered no support in any other form for this phantom study.
Overranging Definition and Components
Two definitions of overranging are in common use: The difference between exposed length and the planned length (which will be used in this article) and the difference between exposed length and imaged length.
The number of overrange rotations, or the corresponding overrange length, is made up of a combination of two effects that occur simultaneously (Fig 1): A limited number of extra rotations are due to the fact that the scanner automatically adds one section width to the user-prescribed planned length because the planned scan positions define the center position of the first and last section to be reconstructed in all scanners. This combination, planned length plus section width, will be termed the imaged length. For example, a planned length of 420 mm with a 5-mm section width in the first reconstruction will generate an imaged length of 425 mm. Another number of extra rotations around the imaged length (425 mm in the example) are necessary to generate sufficient raw data for reconstruction of the first and last sections. These extra rotations do not provide images at table positions that are exposed outside the imaged length. In combination with imaged length, this represents the exposed length.
View larger version (28K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1: Simplified depiction of overranging components and definitions at helical CT scanning. To the planned scan length, one section width (SW) is automatically added so imaged scan length is slightly longer. Extra rotations needed for image reconstruction are added to imaged length, resulting in longer exposed scan length. Definitions of overranging vary; either the difference between planned and exposed scan length (Def 1) or the difference between imaged and exposed scan length (Def 2) is used.
|
|
Although the terms scan range and scan length are both used in practice, to avoid confusion with the term overranging, we will refer to overrange rotations and overrange length only.
CT Scanners Examined
Between October 2003 and June 2004, the following four 16-section CT systems from different vendors were evaluated in this study: (a) a LightSpeed 16 scanner with version 4.1 software (GE Healthcare, Waukesha, Wis), (b) an MX8000IDT scanner with version 2.5 software (Philips Medical Systems, Best, the Netherlands), (c) a Sensation 16 scanner with version VA60B software (Siemens Medical Solutions, Forchheim, Germany), and (d) an Aquilion 16 scanner with version 1.4 software (Toshiba Medical Systems, Otawara, Japan).
Technique for Measuring Overranging Rotations
Free-in-air dose measurements were made by using a 102-mm pencil ionization chamber (model CP-4C; Capintec, Ramsey, NJ) connected to a dosimeter (model 35050A; Keithley Instruments, Cleveland, Ohio). The ionization chamber was fixed to a special stand and was positioned along the central axis of the scanner so that the axis of rotation of the scanner coincided with the center of the ionization chamber. All table movements occurred completely outside the scanner gantry and did not interfere with dose readings (Fig 2).
View larger version (12K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2: Overview of measurement setup. Ionization chamber (I) was fixed to a special stand positioned at rear of scanner and was aligned with gantry axis of rotation. All table movements occurred outside the gantry area. C = collimator, D = detector, Do = dosimeter, F = focal spot.
|
|
In our first experiment, doses were measured at four planned scan lengths for all available collimations and at three to four pitch settings. Planned scan length (PSL) was converted to number of rotations (NR) with the following equation: NR = PSL/TF, where TF is table feed per rotation. All measurements were performed by both authors using a tube rotation time of 0.5 second, a tube voltage of 120 kVp, and a tube charge of 100 mAs. Reconstructed section width was set at 5 mm except for the Philips MX8000IDT scanner in 16 x 0.75-mm mode, where a 3-mm section width was the maximum available value.
Linear regression of measured dose as a function of the number of planned rotations yields, as the y-intercept, the dose at zero planned rotations that can be attributed to overranging (Fig 3). The number of overrange rotations was determined by one author (A.J.v.d.M.) by dividing this y-intercept (dose at zero rotations) by the slope of the graph (dose per rotation). In addition, measured values for overrange rotations were converted to overrange lengths by multiplying by table feed per rotation.
View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3: Representative graph of relationship of dose to planned length (expressed in number of tube rotations, where number of rotations equals planned length divided by table feed). Dose increases linearly with number of rotations. A linear regression technique yields dose per rotation (a), dose from overranging (b), and number of overrange rotations (b/a).
|
|
In a second experiment, for evaluation of the effect of section width, reconstructed section width was varied from the minimum to the maximum selectable width (three to 11 values, depending on the scanner) for all available scan collimation settings and a fixed planned length. All measurements were performed by both authors at a pitch of 1.00, tube rotation time of 0.5 second, tube voltage of 120 kVp, and tube charge of 100 mAs. Dose measurements were converted to number of rotations by dividing them by the slope (dose per rotation) that was derived from linear regression in the first experiment for the given measurement conditions. Overrange rotations equaled the total number of rotations minus the number of rotations needed for the planned scan length.
Evaluation of Organ and Effective Dose at Chest and Abdominal CT
Our data were applied to dose-optimized protocol suggestions for single-phase chest and abdominopelvic CT. Dose optimized implied a balance between diagnostic image quality and dose reduction, not the ultimate in low-dose technique. It should be noted that relative increases in dose through overranging do not depend on whether low-, standard-, or high-dose protocols are used, as long as the same pitch and collimation are employed. These protocol suggestions were supplied by representatives of the various vendors on the basis of cooperation with their clinical partners. The planned length was set at 270 mm for chest CT and 420 mm for abdominal CT (Table 1).
The thyroid and testicles were selected as radiosensitive indicator organs at the borders of the planned scans for chest and abdominal CT, respectively. Organ doses and overall effective doses were calculated for male patients by using a dosimetry calculator (ImPACT CT Dosimetry Calculator, version 0.99W; ImPACT, London, England). To evaluate relative contributions to the doses by the overrange length, two separate dose calculations were performed with (a) the planned length of 270 mm for chest CT or 420 mm for abdominal CT and (b) the total exposed length, including the added length from the measured overrange rotations in the specific scanning conditions. The relative contribution of overranging to organ and effective doses (or the underestimation of the true dose when overranging is neglected) was calculated.
In clinical practice with a patient moving through the gantry, the actual dose profile, as a function of table position, will have a trapezoidal shape owing to convolution effects of tube current, beam width, and table movement (1). Tools available for calculation of patient dose in CT, such as the ImPACT calculator, assume a rectangular dose profile. To evaluate the effect of the shape of the dose profile on patient dose, a trapezoidal profile with a combined rectangular-triangular shape (1) was modeled by modifying a look-up table of the ImPACT calculator (J.G.), with incremental weight factors for the tails of the profile. Variations between rectangular and trapezoidal dose profiles were calculated as percentages.
Statistical Analysis
Because this was a descriptive study on overranging that did not focus on scanner comparisons, statistical analysis focused on correlation of overranging rotations and lengths with scan collimation, pitch, and section width and on differences in doses with and without overranging. Pearson correlation coefficients for collimation, pitch, and section width (with the rectangular dose profile) were calculated by using the data in Tables 2–4 for each scanner separately. Pitch and section width correlations were performed for each collimation separately. Dose differences in Table 5 were evaluated with paired t tests. Calculations were performed with software (SPSS, version 12.0.1, 2003; SPSS, Chicago, Ill). P < .05 was considered to indicate a significant difference. Another software program (CIA, version 1.0, 1989; British Medical Association, London, England) was used to calculate 95% confidence intervals (CIs) for correlations (except for pitch, for which data per collimation were too limited).
View this table:
[in this window]
[in a new window]
|
Table 3. Overrange Length in Millimeters according to Pitch and Scan Collimation for Four CT Scanners and a Rectangular or Trapezoidal Dose Profile
|
|
View this table:
[in this window]
[in a new window]
|
Table 4. Number of Overrange Rotations according to Section Width and Scan Collimation for Four CT Scanners at Constant Planned Length and Pitch of 1.00
|
|
View this table:
[in this window]
[in a new window]
|
Table 5. Thyroid and Testicular Organ Doses, Effective Doses, and Relative Contributions of Overranging for Four CT Scanners and a Rectangular or Trapezoidal Dose Profile
|
|
|
RESULTS
|
|---|
Dose
Dose measurements with increasing planned length expressed in number of tube rotations showed an almost perfect linear relationship for all scanners (Fig 3).
Overrange Rotations
The number of overrange rotations ranged from 1.99 to 4.04 at the lowest pitches to 0.93 to 2.59 at the highest pitches (Table 2). The number of overrange rotations showed a negative correlation with pitch (r values, –0.92 to –0.99). Collimation correlated with the number of overrange rotations only for one scanner (r = –0.75; 95% CI: –0.95, –0.18). When the number of rotations was converted to lengths (Table 3, Fig 4), we found that overrange length showed a positive correlation with pitch (r values, 0.95–1.00) and collimation (r values, 0.84–0.95; 95% CIs: 0.09, 0.98 to 0.74, 0.99) because of the high table feed at increased pitch. Overrange lengths were constant only for the Philips scanner at 0.75-mm scan collimation; this may be due to differences in the reconstruction algorithm between 0.75-mm and 1.5-mm collimation.
View larger version (20K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4: Graph shows relationship between overrange length and pitch at collimations between 1.0 and 1.5 mm for the four CT scanners evaluated (collimation differs per scanner). Reconstructed section width for primary image reconstruction was 5 mm.
|
|
Reconstruction Parameters
The choice of reconstruction parameter section width had a variable effect on the number of overrange rotations (Table 4, Fig 5). With increasing section width, the number of overrange rotations increased linearly (minimum r = 0.91 and 95% CI: 0.56, 0.98 for Siemens scanner; minimum r = 1.00 and 95% CI: 0.87, 1.00 for Toshiba scanner), decreased linearly (minimum r = –0.99 and 95% CI: –1.00, –0.84 for Philips scanner) or showed a nonlinear increasing relationship (minimum r = 0.06 and 95% CI: –0.87, 0.89 for GE Healthcare scanner). The sharp increase in the number of rotations for the Siemens scanner was due to the fact that two reconstruction modes were used: (a) a slower cone-beam–corrected mode at a smaller section width (compared with scan collimation) in which the number of overrange rotations was independent of section width and (b) a faster z-filtering algorithm for a larger section width in which there was a gradual increase in the number of overrange rotations with increasing section width.
View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5: Graph shows relationship between overrange rotations and reconstructed section width (SW) of the primary image reconstruction for the four scanners evaluated. Scan collimation was 1.0–1.5 mm (collimation differs per scanner), pitch was 1.00, and planned length was held constant.
|
|
Radiosensitive Indicator Organs
Because the thyroid and the testicles lie at the borders of the planned length for chest or abdominal CT, overrange rotations have a major impact on the organ doses of these organs (Table 5). The relative contribution of overranging to the organ dose (or the underestimation of the true dose when overranging is neglected) ranged from 59% (0.17/0.29) to 77% (0.34/0.44) for the thyroid and from 40% (0.31/0.78) to 69% (0.9/1.3) for the testicles. Overrange contributions to effective dose ranged from 11% (0.5/4.5) to 17% (0.7/4.1) for chest CT and from 9% (0.7/7.5) to 19% (1.4/7.5) for abdominal CT. Variation is explained by the specific scanner-dependent protocol parameter choices. The higher overrange contribution to the indicator organ dose in the chest protocols is due to the closer proximity of the thyroid to the planned length of a chest CT scan than that of the testicles to the planned length of an abdominal CT scan. Effects on effective dose were less pronounced because the International Commission on Radiological Protection weight factors of thyroid and testicles differ. When planned length was compared with exposed length, the thyroid dose (P = .012), testicular dose (P = .025), effective dose at chest CT (P = .005), and effective dose at abdominal CT (P = .011) were found to be significantly different.
Increases in overrange length from the rectangular to the trapezoidal dose profile measured at the base of the profile were always equal to the beam width. These increases ranged from 21.2% to 47.2%, depending on scanner, collimation, and pitch (Table 3). Using the modified ImPACT calculator, we found that with the trapezoidal dose profile, thyroid doses increased by 14%–34% but testicular doses decreased by 14%–24%. Effective doses were identical for chest CT and decreased 1%–3% for abdominal CT with the trapezoidal profile (Table 5).
|
DISCUSSION
|
|---|
To the best of our knowledge, our evaluation is the first comparative quantification of overranging for frequently used multisection CT scanners. In single-section helical scanning, two classes of image reconstruction algorithms are generally used: linear interpolation with 360° scan data and linear interpolation with 180° scan data. While 360° linear interpolation uses two times 360° of data for image reconstruction, 180° linear interpolation algorithms access a data range of two times 180° (2). In multisection CT, the image reconstruction algorithms are more complex and involve either z-filtering (with four-section systems [4,5]) or cone-beam algorithms (with 16-section systems [6]).
The planned length is what the CT user requests as information from the CT study. For this reason we defined both the extra section width (even though it generates image information) and the additional rotations as components of overranging. For overrange dose measurement, the shape of the dose profile played no role because the measured total dose outside the planned length represented by the area under the up- and downslope tails of a convoluted trapezoidal dose profile equaled the area of the overrange rectangles (derived from the dose at a planned length of zero rotations divided by the dose per rotation) in a rectangular profile. In a rectangular profile, the calculated number of overrange rotations and overrange lengths reflect theoretic values at a full dose that yield an equivalent dose-length product effect as an up- and downsloping dose over a larger length. This may in part explain the lower overrange contributions observed in our study than in the study by Tzedakis et al (1).
In practice, dose profiles will have a trapezoidal shape, leading to larger overrange lengths (by a factor equal to the beam width) but lengths that are exposed at a reduced dose. This will have a bearing on doses to organs that are closer to or further away from the planned scan boundaries. We showed that while thyroid doses increased, testicle doses decreased. Effective doses in body CT were relatively insensitive to dose profile.
In this study, we measured the additional dose from overrange rotations in relation to common imaging parameters—that is, scan collimation, pitch, and reconstructed section width. We found variable numbers of overrange rotations for comparable scan collimations between scanners. With an increase in pitch, the number of overrange rotations decreases, with larger decreases in low pitch ranges than in high pitch ranges. At higher pitches, the decreased number of rotations does not outweigh the higher table feeds, so for all scanners, overrange length increased with increasing pitch and collimation. A small (0.5–0.75-mm) collimation with a high pitch had a more favorable overrange length than did a larger (1.0–2.0-mm) collimation with a lower pitch. A large variability was also demonstrated as an effect of increasing reconstructed section width for a constant planned length, supporting the conclusion that overranging is scanner specific.
In the protocol analysis, overrange lengths varied by a factor of 1.5–2.0 between scanners, depending on parameter choices. Because overranging is independent of planned length, the influence of overranging on effective dose is more important in fast scanning or as planned lengths shorten. The effects on organ dose depend on the overrange length of the protocol and the proximity of the evaluated organ to the scan boundary. Therefore, the contribution from overranging to thyroid dose in chest CT was higher than the contribution to testicular dose in abdominal CT. Differences in contributions to effective dose were less pronounced, owing to different International Commission on Radiological Protection weight factors.
Recently, Tzedakis et al (1) published results of a study on overranging with a Siemens 16-section scanner. Their values for overexposed length for 0.75-mm collimation were similar to ours, but their values for 1.5-mm collimation were somewhat higher than ours. Unfortunately, Tzedakis et al did not specify which data from the console were used to calculate overrange length. Trends in their data closely matched our results, showing larger overrange lengths with increasing pitch and beam collimation and the same characteristic curve for the effect of section width. Using Monte Carlo simulations, they showed that a trapezoidal dose profile already reaches its plateau before and after the boundaries of the planned length. Their values for contributions of overranging to effective dose were higher than our values for the Siemens scanner, probably because of their higher overrange lengths with 1.5-mm collimation. Nicholson and Fetherston (3) described overranging in a Philips four-section unit. They found diminishing overrange lengths with decreases in scan collimation, as well as increasing overrange lengths with increases in pitch. However, overranging was not measured for all combinations of collimation and pitch. Their results are similar to our results with the 16-section Philips scanner.
Effective dose calculations are usually performed with a calculation program or by converting dose-length product to effective dose with conversion factors (7), while the calculation of organ dose requires calculation programs. The appropriate scan length needs to be defined by the user of the program, but overrange lengths are frequently not taken into account, leading to substantial dose underestimations. Thus, insight into overranging is required for proper dose estimations.
Overranging should be taken into account when one is defining CT protocols, especially for short planned lengths and fast scanning (with large collimation and pitch). An unfavorable selection of parameter settings may lead to unnecessary overranging. Overrange length can be minimized by scanning at the thinnest collimation and low pitch (low table feed), while section width effects can be minimized by selecting a minimum section width for the first reconstruction and creating thick sections with thin-slab multiplanar reformation (8). Also, technicians should set the boundaries during scan prescription judiciously, a fact commonly overlooked in multisection scanning (9). Overranging will gain more importance with newer 64-section scanners. These scanners have increased detector widths and enable higher table feeds, leading to longer overrange lengths. This effect may be more important than the slightly improved dose efficiency in the imaged area that is achievable with 64-section scanners as compared with 16-section scanners. Future comparative studies will be needed to define these relative contributions.
Our study had limitations. First, only one CT scanner (with a specific software version) from each vendor was evaluated in detail, and variations between scanners from the same vendor or differences between software versions were not evaluated. Second, we did not separately compare exposed lengths with imaged lengths (planned length plus one section width). We believe that our data from the second experiment allow corrections to be made when a section width other than 5 mm is used in the first reconstruction.
In conclusion, overranging depends on the selection of CT scanning parameters, is scanner specific, and can contribute substantially to doses in sensitive organs at the borders of the planned scan and thus to effective dose. With increases in collimation and pitch, overrange length increases. The relative effect is high with short scan lengths or when a large collimation and high pitch are combined. Therefore, overranging should be included in the calculation of organ and effective doses, and knowledge of the precise behavior of a particular scanner is essential for optimizing scanning protocols with regard to dose-efficient scanning in light of overranging.
|
ADVANCES IN KNOWLEDGE
|
|---|
- Our study yielded estimates of relative contributions of overranging to doses of some radiosensitive organs and to effective doses in CT of the chest and abdomen.
- Our results reinforce the importance of adequate scan length description, including overranging, in dose calculations.
|
ACKNOWLEDGMENTS
|
|---|
We thank Esther van Schrojenstein Lantman, MSc, PDeng, medical physicist in training at the Elkerliek Hospital in Helmond, the Netherlands, and Jannie van den Tillaart, MSc, PDeng, medical physicist in training at the Maxima Medical Center in Veldhoven, the Netherlands for their help in the acquisition of the data.
|
FOOTNOTES
|
|---|
Abbreviations: CI = confidence interval
Author contributions: Guarantors of integrity of entire study, A.J.v.d.M., J.G.; study concepts/study design or data acquisition or data analysis/interpretation, A.J.v.d.M., J.G.; manuscript drafting or manuscript revision for important intellectual content, A.J.v.d.M., J.G.; approval of final version of submitted manuscript, A.J.v.d.M., J.G.; literature research, A.J.v.d.M.; experimental studies, A.J.v.d.M., J.G.; and manuscript editing, A.J.v.d.M., J.G.
Authors stated no financial relationship to disclose.
|
References
|
|---|
- Tzedakis A, Damilakis J, Perisinakis K, Stratakis J, Gourtsoyiannis N. The effect of overscanning on patient effective dose from multidetector helical computed tomography examinations. Med Phys 2005;32:1621–1629.[CrossRef][Medline]
- Kalender WA. Computed tomography: fundamentals, system technology, image quality, applications. 2nd revised ed. Erlangen, Germany: Publicis Corporate, 2005; 82–87.
- Nicholson R, Fetherston S. Primary radiation outside the imaged volume of a multislice helical CT scan. Br J Radiol 2002;75:518–522.[Abstract/Free Full Text]
- Taguchi K, Aradate H. Algorithm for image reconstruction in multislice helical CT. Med Phys 1998;25:550–561.[CrossRef][Medline]
- Schaller S, Flohr T, Klingenbeck K, Krause J, Fuchs T, Kalender WA. Spiral interpolation algorithm for multislice spiral CT. I. Theory. IEEE Trans Med Imaging 2000;19:822–834.
- Flohr T, Stierstorfer K, Bruder H, Simon J, Polacin A, Schaller S. Image reconstruction and image quality evaluation for a 16-slice CT scanner. Med Phys 2003;30:832–845.[CrossRef][Medline]
- Bongartz G, Golding SJ, Jurik AG, et al. European guidelines for multislice computed tomography. March 2004. http://www.msct.info/CT_Quality_Criteria.htm. Accessed October 2, 2006.
- Prokop M. Principles of CT, spiral CT and multislice CT. In: Prokop M, Galanski M, van der Molen AJ, Schaefer-Prokop CM, eds. Spiral and multislice computed tomography of the body. 2nd ed. Stuttgart, Germany: Thieme Verlag, 2003; 1–43.
- Kalra MK, Maher MM, Toth TL, Kamath RS, Halpern EF, Saini S. Radiation from "extra" images acquired with abdominal and/or pelvic CT: effect of automated tube current modulation. Radiology 2004;232:409–414.[Abstract/Free Full Text]
TOP
ABSTRACT
INTRODUCTION
