Published online before print February 9, 2007, 10.1148/radiol.2431060696
(Radiology 2007;243:96-104.)
© RSNA, 2007
Dosimetry and Adequacy of CT-based Attenuation Correction for Pediatric PET: Phantom Study1
Frederic H. Fahey, DSc,
Matthew R. Palmer, PhD,
Keith J. Strauss, MS,
Robert E. Zimmerman, MSEE,
Ramsey D. Badawi, PhD and
S. Ted Treves, MD
1 From the Division of Nuclear Medicine and Department of Radiology, Children's Hospital Boston, 300 Longwood Ave, Boston, MA 02115 (F.H.F., K.J.S., S.T.T.); Division of Nuclear Medicine, Beth Israel Deaconess Medical Center, Boston, Mass (M.R.P.); Division of Nuclear Medicine, Brigham and Women's Hospital, Boston, Mass (R.E.Z.); and Department of Radiology, University of California at Davis, Davis, Calif (R.D.B.). Received April 21, 2006; revision requested June 23; revision received July 12; accepted July 21; final version accepted September 1.
Address correspondence to F.H.F. (e-mail: frederic.fahey{at}childrens.harvard.edu).
 |
ABSTRACT
|
|---|
Purpose: To evaluate the dose from the computed tomographic (CT) portion of positron emission tomography (PET)/CT to determine minimum CT acquisition parameters that provide adequate attenuation correction.
Materials and Methods: Measurements were made with a PET/CT scanner or a PET scanner, five anthropomorphic phantoms (newborn to medium adult), and an ionization chamber. The CT dose was evaluated for acquisition parameters (10, 20, 40, 80, 160 mA; 80, 100, 120, 140 kVp; 0.5 and 0.8 second per rotation; 1.5:1 pitch). Thermoluminescent dosimetry was used to evaluate the germanium 68/gallium 68 rod sources. A phantom study was performed to evaluate CT image noise and the adequacy of PET attenuation correction as a function of CT acquisition parameters and patient size.
Results: The volumetric anthropomorphic CT dose index varied by two orders of magnitude for each phantom over the range of acquisition parameters (0.30 and 21.0 mGy for a 10-year-old with 80 kVp, 10 mAs, and 0.8 second and with 140 kVp, 160 mAs, and 0.8 second, respectively). The volumetric anthropomorphic CT dose index for newborn phantoms was twice that for adult phantoms acquired similarly. The rod source dose was 0.03 mGy (3-minute scan). Although CT noise varied substantially among acquisition parameters, its contribution to PET noise was minimal and yielded only a 2% variation in PET noise. In a pediatric phantom, PET images generated by using CT performed with 80 kVp and 5 mAs for attenuation correction were visually indistinguishable from those generated by using CT performed with 140 kVp and 128 mAs. With very-low-dose CT (80 kVp, 5 mAs) for the adult phantom, undercorrection of the PET data resulted.
Conclusion: For pediatric patients, adequate attenuation correction can be obtained with very-low-dose CT (80 kVp, 5 mAs, 1.5:1 pitch), and such correction leads to a 100-fold dose reduction relative to diagnostic CT. For adults undergoing CT with 5 mAs and 1.5:1 pitch, the tube voltage needs to be increased to 120 kVp to prevent undercorrection.
© RSNA, 2007
The clinical use of hybrid positron emission tomographic (PET)/computed tomographic (CT) scanners has grown substantially in the past few years, and such growth has led to an increase in the use of CT-based attenuation correction (1–9). Conventionally, attenuation correction factors for PET are determined by using positron-emitting rod sources that rotate about the patient. The collection of the transmission data from these rod sources may take 10–20 minutes, whereas the CT scan can be obtained in less than 1 minute. This reduction of image time may be of particular interest in pediatric imaging. Also, a CT scan uses substantially more photons than do the rotating rod sources, and, thus, less noise is introduced into the subsequent PET reconstruction. In CT-based attenuation correction, CT data are acquired over the region of the body to be scanned with PET and are reconstructed into a series of transverse CT images. A multilinear transformation is applied to estimate the linear attenuation coefficient at 511 keV (µ511) from the CT numbers (Hounsfield units) (10). After the transformation is applied, these data are smoothed to a degree comparable to the resolution of PET to minimize edge artifacts. The resulting transformed and smoothed images are then forward-projected to generate the CT-based attenuation correction file, which is applied to the emission data.
There has been concern about the radiation dose from CT, particularly for the pediatric age group. Although only 4% of all imaging procedures involve CT, these procedures lead to 40% of the radiation dose commitment delivered to the US population from radiologic procedures (11). The effective dose from diagnostic abdominal CT is estimated to be in the range of 5–10 mSv and is even higher for smaller patients (12). When CT is applied as part of PET/CT, the axial field of view is much greater than that of traditional abdominal CT and routinely extends from the base of the skull to the proximal thigh; the effective dose thus may be two to four times greater. This dose can be compared with the effective dose from fluorine 18 (18F) fluorodeoxyglucose PET of approximately 10 mSv for administration of 520 MBq of 18F fluorodeoxyglucose (13). Thus, the purpose of our study was to evaluate the dose from the CT portion of PET/CT to determine the minimum CT acquisition parameters that provide adequate attenuation correction.
|
MATERIALS AND METHODS
|
|---|
Dosimetry
Measurements were made by using a PET/CT scanner (Discovery LS; GE Healthcare, Milwaukee, Wis) or a PET scanner (Advance NXi; GE Healthcare). The PET component of the PET/CT scanner is the same as that of the PET scanner, and thus the two devices were considered equivalent in this study. Because of scanner availability, measurements that required only the PET component of the scanner were made with the PET scanner. The CT portion of the PET/CT scanner is a four-section scanner. The measurements were made by using a series of tissue-equivalent abdominal phantoms (CIRS, Norfolk, Va) designed for use with CT. The phantoms are made of an epoxy formulation that mimics the absorption and scatter characteristics of tissue to within 1% in the diagnostic range of x-ray energies. There is also a simulated vertebral bone component in the posterior portion of the phantom. Five 1.3-cm-diameter holes located in the center, anterior, posterior, right lateral, and left lateral positions allow the placement of dosimeters within the phantom (Table 1).
The dosimetric measurements were estimated by using a CT pencil ionization chamber (Inovision 500-100; Inovision Radiation Measurements, Cleveland, Ohio), with 3-mL volume and 10-cm active length. These dosimetric measurements were performed by three authors (F.H.F., M.R.P., K.J.S.). Data were acquired for five phantoms (newborn, 1-year-old, 5-year-old, 10-year-old, and medium adult phantoms) at tube voltages of 80, 100, 120, and 140 kVp. These five phantoms were used because they represent a reasonable range of patient sizes for children and adults. Except where indicated, a tube current of 160 mA and an exposure time of 0.8 second (128 mAs) were used. Four rows of detectors with a section thickness of 2.5 mm per detector for a 10-mm collimated fan beam were used with a table speed of 15 mm per rotation for a 1.5:1 pitch. For each phantom and CT setting, measurements were made with the ionization chamber placed sequentially in the center, anterior, posterior, and left lateral positions. Each measurement was repeated three times, and the average was reported.
To test linearity of certain parameters, additional data were acquired by using the medium adult phantom with the ionization chamber in the center position at 80 kVp and 0.8 second per rotation. Tube current linearity was evaluated by acquiring data with three tube currents (10, 160, and 320 mA), and the exposure time linearity was evaluated by using three exposure times (0.5, 0.8, and 2.0 seconds). The effect of section thickness on exposure was evaluated at three section thicknesses (5, 10, and 20 mm).
CT dose indexes were calculated for the anthropomorphic phantoms by using a series of relationships developed for the standard CT dose index phantom (14). Since we used anthropomorphic phantoms rather than the standard round plastic (Plexiglas) phantoms, we designated our value as anthropomorphic CT dose index, or CTADI, to distinguish it from the standard definitions of CT dose index. The anthropomorphic CT dose index for a beam width of 100 mm, or CTADI100, was calculated with the following equation:
where X is the exposure measured by the ionization chamber in roentgens; 8.76 is the ratio between dose in air, or Dair (in milligrays), and exposure, or Exp (in roentgens); f(Mat/Air) corrects for the material of interest relative to air; and Nomwidth is nominal fan beam width. The last term in the equation scales the results to a nominal fan beam width of 100 mm. Weighted anthropomorphic CT dose index, or CTADIw, was calculated as follows: CTADIw = (2/3)CTADI100e + (1/3)CTADI100c, where CTADI100e is the average anthropomorphic CT dose index for a beam width of 100 mm on the edge of the patient and CTADI100c is the value at the center. The volumetric anthropomorphic CT dose index, or CTADIvol, was calculated as follows: CTADIvol = CTADIw · (NT/I) = CTADIw · 1/P, where N is the number of detectors, T is the section thickness per detector (NT is the beam collimation), I is the table displacement per rotation, and P is pitch. Alternatively, pitch is the table displacement per rotation divided by the beam collimation. The volumetric anthropomorphic CT dose index was calculated and tabulated for each phantom and set of CT acquisition parameters.
Thermoluminescent dosimetry was used to evaluate the dose associated with the use of the rotating rod sources because the exposure rate provided by the rotating rod sources was insufficient for recording with the ionization chamber that was used for the CT measurements. At the time of these measurements, the total activity of the two sources was 540 MBq of germanium 68/gallium 68. Two 3.1 x 3.1-mm and 0.89-mm-thick lithium fluoride dosimeter chips were placed at each position within the five phantoms. Each phantom was scanned for 1 hour. Subsequently, the chips were processed, and the results were reported to the nearest 0.01 mGy. The average of the readings of the two chips at each location was reported. From these data, a weighted dose index, or DIw, analogous to the weighted CT dose index, was calculated. In other words, DIw = (2/3)De + (1/3)Dc, where De is the average radiation dose on the edge of the phantom and Dc is the average radiation dose in the center reported in milligrays. This result was subsequently scaled for a 3-minute acquisition because this time is a reasonable duration for clinical imaging.
Noise Evaluation
Two additional phantoms to which radioactivity could be added were used for these experiments. The cylindric phantom defined by the 1994 document of the National Electrical Manufacturers Association (15) (20-cm diameter and length) was used and is referred to hereafter as the NU2-94 phantom. This phantom is similar in size to the 10-year-old phantom in our anthropomorphic set and, thus, is considered the size of a child. The second phantom was a torso phantom (Data Spectrum, Hillsborough, NC). It has an anteroposterior thickness of 26 cm and a lateral thickness of 38 cm and contains two lung compartments (filled with a combination of small expanded rigid polystyrene plastic [Styrofoam] balls and water), a liver compartment, and a polytetrafluoroethylene (Teflon) spine insert. This phantom is considered to be the size of a medium to large adult relative to our anthropomorphic set.
The NU2-94 phantom was imaged with the background portion filled with 56.6 MBq (10.4 kBq/mL) of 18F. A 6-hour three-dimensional emission scan (2.5 x 109 true coincidences) was acquired. A long emission scan was acquired to minimize its contribution to the reconstructed noise, such that the contribution from the CT-based attenuation correction could be evaluated. All CT scans were acquired with a rotation speed of 0.8 second per rotation, four rows of detectors with a section thickness of 2.5 mm per detector (10-mm beam collimation), and a 1.5:1 pitch. Twenty CT scans were acquired with five tube currents (10, 20, 40, 80, 160 mA) and four tube voltages (80, 100, 120, 140 kVp). The same PET emission data were reconstructed 20 separate times by using each of the acquired CT scans for attenuation correction and Fourier rebinning, followed by application of an ordered subset expectation maximization, or OSEM, algorithm (28 subsets, two iterations) with a postreconstruction filter of 4.36 mm and a loop filter of 3.12 mm. These parameters were used because they are routinely used to reconstruct our pediatric whole-body PET scans. Five circular regions of interest (ROIs) numbered from one to five that were all the same size (approximately 4 cm in diameter) were randomly placed in the background region of the phantom by one of the authors (F.H.F.). For each ROI, the mean and standard deviation of the counts within the ROI were compared for each tube voltage and current combination. The magnitude of the noise was characterized by the coefficient of variation (ie, the ROI standard deviation divided by the ROI mean and represented as a percentage). The same five ROIs were applied to the reconstructed CT scans to compare the CT and PET noise values.
Both the NU2-94 and the torso phantoms were used to further evaluate the contribution of CT noise to the overall noise on the reconstructed PET scan. In this experiment, the NU2-94 phantom was imaged in the two-dimensional mode. The torso phantom was filled with 481 MBq (44 kBq/mL) of 18F, and the NU2-94 phantom was filled with 60 MBq (7.4 kBq/mL) of 18F. A single 6-hour three-dimensional emission scan was acquired for each phantom. Totals of 1.1 x 108 and 8.6 x 108 true coincidences were acquired for the NU2-94 and torso phantoms, respectively. CT scans of the two phantoms were acquired with a rotation speed of either 0.5 or 0.8 second per rotation. Four rows of detectors with a section thickness of 2.5 mm per detector (10-mm beam collimation) were used with a pitch of 1.5:1. CT data were acquired with rotation speeds of 0.5 and 0.8 second per rotation; tube currents of 10, 20, 40, 80, and 160 mA; and tube voltages of 80, 100, 120, and 140 kVp. For selected tube current, tube voltage, and rotation speed combinations (Table 2), 10 separate CT scan realizations were acquired and used to reconstruct the PET emission data. In each case, the mean and standard deviation of each pixel across the 10 CT realizations were determined, and the coefficient of variation was calculated. In this way, the noise contribution from the CT-based attenuation correction to the PET reconstruction was determined.
Adequacy of Attenuation Correction
The reconstructed images of the NU2-94 and torso phantoms were subjectively evaluated with consensus of two authors (F.H.F. and M.R.P., each with 3 years of PET/CT experience) for the adequacy of CT-based attenuation correction. Attenuation correction was considered adequate if the resultant reconstructed PET data were visually indistinguishable from the data that were reconstructed with the routine diagnostic-quality CT scan. In addition, the accuracy of the estimation of µ511 by using CT-based attenuation correction was evaluated as a function of tube current, tube voltage, rotation speed, and patient size. For this evaluation, no emission data were necessary, and thus the entire set of anthropomorphic CT phantoms in Table 1 was used. CT scans were acquired for each phantom by using tube current, tube voltage, and rotation speed combinations for rotation speeds of 0.5 and 0.8 second per rotation; tube currents of 10 and 160 mA; and tube voltages of 80, 100, 120, and 140 kVp. All CT scans were acquired with a 1.5:1 pitch. These data were processed as if they were to be used for PET attenuation correction. The attenuation correction matrix was reconstructed with a ramp filter and provided an image in which each pixel value represented an estimate of µ511 at that location. In each case, a circular ROI was drawn in the central uniform portion of the phantom. ROIs on all images were the same size, approximately 5 cm in diameter. The mean and standard deviation of the CT pixel values within the ROI were recorded. The ROIs were drawn, and the data were recorded by one of the authors (F.H.F.).
|
RESULTS
|
|---|
Dosimetry
Obtaining a CT scan of a child by using the same acquisition parameters as those used for obtaining a CT scan in an adult (Table 3) will lead to a substantial increase in radiation dose (increases of 94%–111%, 72%–83%, 60%–65%, and 44%–48% for the newborn, 1-year-old, 5-year-old, and 10-year-old phantoms, respectively). This increase in radiation dose to smaller patients is caused by less attenuation, and, thereby, the center of the smaller patient receives a higher dose. For the same patient, reducing the tube voltage and current from 140 kVp and 160 mA to 80 kVp and 10 mA leads to a reduction in the radiation dose by a factor of between 66 and 73. If the rotation speed is also increased from 0.8 to 0.5 second per rotation, then the dose reduction factor is more than 100.
Values of weighted dose index for the rotating rod sources, scaled to a 3-minute transmission acquisition, were 0.0354, 0.038, 0.0328, 0.0321, and 0.0316 mGy for the newborn, 1-year-old, 5-year-old, 10-year-old, and medium adult phantoms, respectively. The increase in radiation dose to children compared with the medium-sized adult is not as great as with CT (12.0%, 7.1%, 3.7%, and 1.6% for newborn, 1-year-old, 5-year-old, and 10-year-old phantoms, respectively). In addition, the radiation dose from the rotating rod sources is substantially less than that from CT, even for the lowest tube voltages and tube currents. For the 10-year-old phantom, the radiation dose from the rotating rod sources was 9.3 times less than that for CT with 80 kVp, 10 mA, 0.8-second rotation, and 1.5:1 pitch and 655 times less than that for CT with 140 kVp, 160 mA, 0.8-second rotation, and 1.5:1 pitch.
Noise Evaluation
The effect of tube current and tube voltage on the noise in the CT scan was evaluated (Fig 1). In regard to the effect of CT noise on the subsequently reconstructed PET scan, there was a slight increase (about 2%) in the PET noise as the tube current was decreased from 80 to 10 mA. Figure 1a shows that this same reduction in current led to a 400% increase in CT noise. Thus, large increases in CT noise did not lead to similar increases in PET noise. In regard to the contribution of CT noise to the PET scan, the coefficient of variation was less than 1% for all CT acquisitions with the NU2-94 phantom. For the torso phantom, the coefficient of variation was 5.64%, 1.65%, and 0.45% for CT acquisitions with 10, 40, and 160 mA, respectively, and 7.65% for 80 kVp, 10 mA, and 0.5 second per rotation (Fig 2).
View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1a: Graphs show CT noise in the NU2-94 phantom in terms of the standard deviation of the CT numbers for five background regions (ROIs 1–5). (a) Graph shows values as a function of tube current, with tube voltage of 80 kVp. (b) Graph shows values as a function of tube voltage, with tube current of 160 mA. In both cases, the rotation speed was 0.8 second per rotation and the pitch was 1.5:1. ROI 4 is in the center of the phantom, and its values are slightly noisier due to photon attenuation.
|
|
View larger version (14K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1b: Graphs show CT noise in the NU2-94 phantom in terms of the standard deviation of the CT numbers for five background regions (ROIs 1–5). (a) Graph shows values as a function of tube current, with tube voltage of 80 kVp. (b) Graph shows values as a function of tube voltage, with tube current of 160 mA. In both cases, the rotation speed was 0.8 second per rotation and the pitch was 1.5:1. ROI 4 is in the center of the phantom, and its values are slightly noisier due to photon attenuation.
|
|
View larger version (32K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2: Graph shows contribution to PET noise from CT-based attenuation correction. All scans except where otherwise indicated were acquired with 80 kVp, 0.8 second per rotation, and 1.5:1 pitch. For the torso phantom, scans also were acquired with 80 kVp, 10 mA, 0.5 second per rotation, and 1.5:1 pitch. The standard deviation of 10 realizations of the CT scan at each set of acquisition parameters was used to determine the CT contribution to the PET noise for the NU2-94 (NEMA NU2) and torso phantoms. For the NU2-94 phantom, the coefficient of variation was less than 1% in all cases. For the torso phantom, the coefficient of variation ranged from less than 1% to 7.65%.
|
|
Adequacy of Attenuation Correction
PET scans of the NU2-94 phantom that were reconstructed by using CT scans with any of the acquisition parameters were essentially identical (Fig 3, C and D); this result indicated that, for a phantom of this size, an excellent reconstruction could be obtained even when a very noisy CT scan was used. However, a similar comparison for the torso phantom indicated that PET image data reconstructed by using very-low-dose CT were substantially undercorrected (Fig 4, C). This effect resulted from the fact that very high statistical noise in the CT data could lead to a bias in the estimate of µ511. The phantoms consisted mainly of water and, consequently, should have Hounsfield unit values of zero and a µ511 of approximately 0.093 cm–1 (Fig 5). With the very-low-dose CT scans, the images were very noisy, with a substantial variation in the Hounsfield unit values of about zero. When these values were applied to the multilinear transformation between Hounsfield units and µ511 (Fig 5a), the Hounsfield unit values that were greater than zero were undervalued relative to those that were less than zero, and an underestimation of µ511 resulted. Therefore, with high noise levels at CT, the value for µ511 was underestimated with values less than 0.090 cm–1; thus, undercorrection in the reconstructed data resulted.
View larger version (51K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3: Adequacy of CT-based attenuation correction in the uniform NU2-94 phantom. A, Transverse CT scan for 80 kVp, 10 mA, 0.5 second per rotation, and 1.5:1 pitch. B, Transverse CT scan for 140 kVp, 160 mA, 0.8 second per rotation, and 1.5:1 pitch. C, Transverse PET scan reconstructed by using the CT scan in A for CT-based attenuation correction. D, Transverse PET scan reconstructed by using the CT scan in B for CT-based attenuation correction. Note that the reconstructed PET data in C and D are essentially identical.
|
|
View larger version (60K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4: Adequacy of CT-based attenuation correction in the torso phantom. A, Transverse CT scan for 80 kVp, 10 mA, 0.5 second per rotation, and 1.5:1 pitch. B, Transverse CT scan for 140 kVp, 160 mA, 0.8 second per rotation, and 1.5:1 pitch. C, Transverse PET scan reconstructed by using the CT scan in A for CT-based attenuation correction. D, Transverse PET scan reconstructed by using the CT scan in B for CT-based attenuation correction. Note that the use of very low–dose CT with the larger phantom leads to undercorrection with CT-based attenuation correction.
|
|
View larger version (10K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a: Effect of CT noise on adequacy of CT-based attenuation correction. (a) Graph shows relationship of CT noise (graph at bottom), CT number, and µ511 transformation. The graph at left shows the transformed distribution of CT numbers. The wide range in CT numbers leads to an underestimation in µ511. (b) Plot of CT noise (standard deviation) versus µ511. CT scans of all eight of the tissue-equivalent CT phantoms were acquired, and these were processed into attenuation correction files and reconstructed to yield estimates of µ511. Both calculated (based on a Gaussian model) and measured data were plotted, and there was very good agreement between the calculated and measured data. The slight bias is most likely a result of the inadequacy of the Gaussian noise assumption.
|
|
View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b: Effect of CT noise on adequacy of CT-based attenuation correction. (a) Graph shows relationship of CT noise (graph at bottom), CT number, and µ511 transformation. The graph at left shows the transformed distribution of CT numbers. The wide range in CT numbers leads to an underestimation in µ511. (b) Plot of CT noise (standard deviation) versus µ511. CT scans of all eight of the tissue-equivalent CT phantoms were acquired, and these were processed into attenuation correction files and reconstructed to yield estimates of µ511. Both calculated (based on a Gaussian model) and measured data were plotted, and there was very good agreement between the calculated and measured data. The slight bias is most likely a result of the inadequacy of the Gaussian noise assumption.
|
|
Figure 6a shows the measured estimate for µ511 in a central ROI for phantoms of different sizes for the CT parameters that yielded the lowest possible dose (80 kVp, 10 mA, 0.5 second per rotation, and 1.5:1 pitch). With all of the pediatric phantoms, an adequate µ511 value (in excess of 0.090 cm–1) was provided, but with all adult phantoms an underestimation (Fig 6a) occurred. For the lowest tube current–time product settings (0.5 second per rotation and 10 mA), 120 kVp is necessary for the large adult phantom (Fig 6b). It should be noted that increasing the tube voltage rather than the tube current–time product to obtain an adequate attenuation correction will lead to an overall lower patient radiation dose.
View larger version (45K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a: Effect of patient size on adequacy of CT-based attenuation correction. (a) Graph shows µ511 versus patient size for 80 kVp, 10 mA, 0.5 second per rotation, and 1.5:1 pitch. An appropriate value for µ511 (>0.0900 cm–1) was obtained for all pediatric phantoms but not for the adult phantoms. (b) Graph shows µ511 versus tube voltage, tube current, and rotation speed for the adult phantoms. All CT data were acquired with 1.5:1 pitch. Note that, for 10 mA, 120 kVp is necessary and that, for 80 kVp, 40 mA would be necessary for adequate CT-based attenuation correction.
|
|
View larger version (36K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b: Effect of patient size on adequacy of CT-based attenuation correction. (a) Graph shows µ511 versus patient size for 80 kVp, 10 mA, 0.5 second per rotation, and 1.5:1 pitch. An appropriate value for µ511 (>0.0900 cm–1) was obtained for all pediatric phantoms but not for the adult phantoms. (b) Graph shows µ511 versus tube voltage, tube current, and rotation speed for the adult phantoms. All CT data were acquired with 1.5:1 pitch. Note that, for 10 mA, 120 kVp is necessary and that, for 80 kVp, 40 mA would be necessary for adequate CT-based attenuation correction.
|
|
|
DISCUSSION
|
|---|
Dosimetry
Wu et al (16) evaluated the dosimetry associated with CT-based attenuation correction by using thermoluminescent dosimetry with an anthropomorphic phantom (Rando Alderson; Phantom Laboratory, Salem, NY). For a whole-body scan, these investigators estimated effective doses of 18.97 mSv (140 kVp, 80 mA, 0.8 second per rotation, 0.75:1 pitch), 8.81 mGy (140 kVp, 80 mA, 0.8 second per rotation, 1.5:1 pitch), and 0.72 mGy (140 kVp, 10 mA, 0.5 second per rotation, 1.5:1 pitch). These values of effective dose are comparable to our measurements of volumetric anthropomorphic CT dose index for our medium adult phantom, but they are approximately 20% higher for the same acquisition parameters. These differences are most likely caused by the difference between the definition of effective dose and volumetric anthropomorphic CT dose index and the differences in the two phantoms used in our study and that of Wu et al. These investigators estimated that the effective dose from the rotation sources for a 35-minute whole-body scan is 0.26 mSv, compared with our value of 0.37 mGy, when it was adjusted for the same duration. Unlike in our investigation, in that of Wu et al, the dosimetry associated with CT-based attenuation correction was not evaluated as a function of patient size.
Adequacy of Attenuation Correction
Kamel et al (17) evaluated the effect of lowering the tube current on the adequacy of CT-based attenuation correction and its effect of tumor quantification as compared with the use of rotating rod sources. They showed that there was no substantial difference in the estimates of 18F fluorodeoxyglucose uptake or tumor size with varying tube current. There was a slight difference in the uptake estimates by using CT-based versus rotating source–based attenuation correction, particularly for tumors with high uptake, but the investigators stated that this difference would not be clinically relevant in most cases. Our findings are consistent with these results in the range of acquisition parameters used. For 140 kVp, our study findings indicated that CT-based attenuation correction is adequate for all possible tube currents. However, for the combination of both lower tube voltage and tube current, we found that CT-based attenuation correction is adequate for pediatric patients but not for adult patients. In our evaluation of the adequacy of the attenuation correction by using low-dose CT, we found that PET scans for larger patients were undercorrected because of the underlying noise in the CT scan.
Study Limitations
These experiments were performed with one type of PET/CT scanner, and, thus, some of the conclusions might not be applicable to PET/CT scanners of other designs. The results of the dosimetry studies will be, in general, applicable to scanners of other designs, but applicability of the results of the noise evaluation and adequacy of attenuation correction will depend on the reconstruction algorithms and data processing approaches used. We presented our dosimetric measurements in a unit described as the anthropomorphic CT dose index because of our use of a specific set of anthropomorphic phantoms. Although the use of these phantoms allowed us to present data that were applicable to pediatric PET/CT, these results may not be directly comparable to reported CT dose index values.
Last, we conclude that very-low-dose CT scans can be used to provide adequate CT-based attenuation correction, but we did not consider the minimum dose of CT that provides adequate anatomic correlation to the PET scan. This was beyond the scope of our investigation.
In summary, very-low-dose CT scans provide adequate CT-based attenuation correction for PET in pediatric patients, whereas the tube voltage must be increased slightly to 120 kVp for adult patients when the lowest tube current–time product setting (5 mAs) is used.
Practical application: Results of our investigation indicate that adequate CT-based attenuation can be applied with very-low-dose CT scans. Therefore, if one is interested in acquiring a CT scan in pediatric PET only for the use of CT-based attenuation correction, the dose from the CT scan can be reduced by as much as a factor of 100 relative to that required for a diagnostic CT scan. However, if one desires to acquire a diagnostic-quality CT scan, this investigation also shows that CT scans, acquired according to the dose reduction schema proposed by Donnelly et al (18) or Huda et al (19) for pediatric patients, can be used to provide adequate CT-based attenuation correction. There may also be some instances in which the acquisition of a low-dose CT scan for attenuation correction may be appropriate for adult PET/CT as well, and these instances include brain PET, in which magnetic resonance imaging is the preferred anatomic modality, or instances in which acquisition of such scans may be used to remedy cases of misalignment between the CT and emission scans. Low-dose CT also may be used in conjunction with stress-rest PET cardiac protocols by using rubidium 82; in these protocols, several emission scans are acquired over time, and it may be necessary to acquire a separate CT scan for attenuation correction for each emission scan.
|
ADVANCES IN KNOWLEDGE
|
|---|
- Adequate attenuation correction can be obtained for pediatric patients with very-low-dose CT, and the use of very-low-dose CT leads to a 100-fold dose reduction relative to diagnostic CT.
- For adults, the tube voltage needs to be slightly increased to prevent undercorrection.
|
ACKNOWLEDGMENTS
|
|---|
The authors thank William Lorenzen, MS, and Kevin Buckley, MSc, of Children's Hospital Boston, Boston, Mass, and Jeffrey English, BS, of Beth Israel Deaconess Medical Center, Boston, Mass, for their helpful contributions and suggestions.
|
FOOTNOTES
|
|---|
Abbreviations: µ511 = linear attenuation coefficient at 511 keV • ROI = region of interest
The wife of R.D.B. holds less than $10 000 in stock from GE Healthcare, manufacturer of the scanner used in this study.
Author contributions: Guarantor of integrity of entire study, F.H.F.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, F.H.F., M.R.P., R.E.Z., S.T.T.; experimental studies, F.H.F., M.R.P., K.J.S., R.E.Z., S.T.T.; statistical analysis, F.H.F.; and manuscript editing, all authors
|
References
|
|---|
- Beyer T, Townsend DW, Brun T, et al. A combined PET/CT scanner for clinical oncology. J Nucl Med 2000;41:1369–1379.[Abstract/Free Full Text]
- Schöder H, Erdi YE, Larson SM, Yeung HW. PET/CT: a new imaging technology in nuclear medicine. Eur J Nucl Med Mol Imaging 2003;30:1419–1437.[CrossRef][Medline]
- Steinert HC, von Schulthess GK. Initial clinical experience using a new integrated in-line PET/CT system. Br J Radiol 2002;75(spec issue):S36–S38.[Free Full Text]
- Wahl RL. Why nearly all PET of abdominal and pelvic cancers will be performed as PET/CT. J Nucl Med 2004;45(suppl 1):82S–95S.[Abstract/Free Full Text]
- Cohade C, Osman M, Leal J, Wahl RL. Direct comparison of (18)F-FDGPET and PET/CT in patients with colorectal carcinoma. J Nucl Med 2003;44:1797–1803.[Abstract/Free Full Text]
- Nakamoto Y, Osman M, Cohade C, et al. PET/CT: comparison of quantitative tracer uptake between germanium and CT transmission attenuation-corrected image. J Nucl Med 2002;43:1137–1143.[Abstract/Free Full Text]
- Kinahan PE, Hasegawa BH, Beyer T. X-ray-based attenuation correction for positron emission tomography/computed tomography scanners. Semin Nucl Med 2003;33:166–179.[CrossRef][Medline]
- Kinahan PE, Townsend DW, Beyer T, Sashin D. Attenuation correction for a combined 3D PET/CT scanner. Med Phys 1998;25:2046–2053.[CrossRef][Medline]
- Burger C, Goerres G, Schoenes S, et al. PET attenuation coefficients from CT images: experimental evaluation of the transformation of CT into PET 511 keV attenuation coefficients. Eur J Nucl Med Mol Imaging 2002;29:922–927.[CrossRef][Medline]
- Lonn AHR. Evaluation of method to minimize the effect of x-ray contrast attenuation correction. In: 2003 IEEE Nuclear Science Symposium: conference record—19-25 October, 2003, Portland, Oregon, USA. Vol 3. Piscataway, NJ: Institute of Electrical and Electronics Engineers, 2004; 2220–2221.
- Hall E. Commentary on Roebuck DJ: "Risk and benefit in paediatric radiology". Pediatr Radiol 1999;29:721–722.[CrossRef][Medline]
- Huda W. Effective dose to adult and pediatric patients. Pediatr Radiol 2002;32:272–279.[CrossRef][Medline]
- International Commission on Radiological Protection. Radiation dose to patients from radiopharmaceuticals. International Commission on Radiological Protection report 80. New York, NY: Pergamon, 1999; 49.
- Shope TB, Gagne RM, Johnson GC. A method for describing the dose delivered by transmission x-ray computed tomography. Med Phys 1981;8:488–495.[CrossRef][Medline]
- National Electrical Manufacturers Association. Performance measurements of positron emission tomographs. Report NU2. Washington, DC: National Electrical Manufacturers Association, 1994.
- Wu TH, Huang YH, Lee JJ, et al. Radiation exposure during transmission measurements: comparison between CT- and germanium-based techniques with a current PET scanner. Eur J Nucl Med Mol Imaging 2004;31:38–43.[CrossRef][Medline]
- Kamel E, Hany TF, Burger C, et al. CT vs 68Ge attenuation correction in a combined PET/CT system: evaluation of the effect of lowering the CT tube current. Eur J Nucl Med Mol Imaging 2002;29:346–350.[CrossRef][Medline]
- Donnelly LF, Emery KH, Brody AS, et al. Minimizing radiation dose for pediatric body applications of single-detector helical CT: strategies at a large children's hospital. AJR Am J Roentgenol 2001;176:303–306.[Free Full Text]
- Huda W, Scalzetti EM, Levin G. Technique factors and image quality as functions of patient weight at abdominal CT. Radiology 2000;217:430–435.[Abstract/Free Full Text]
TOP
ABSTRACT
MATERIALS AND METHODS
