- Email: [email protected]
The Dosimetric Effect of Zipper Artifacts on TomoTherapy Adaptive Dose Calculation—A Phantom Study
Medical Dosimetry, Vol. 36, No. 3, pp. 306-312, 2011 Copyright © 2011 American Association of Medical Dosimetrists Printed in the USA. All rights reserved 0958-3947/11/$–see front matter
doi:10.1016/j.meddos.2010.06.002
THE DOSIMETRIC EFFECT OF ZIPPER ARTIFACTS ON TOMOTHERAPY ADAPTIVE DOSE CALCULATION—A PHANTOM STUDY WING-KEI
HUI GENG, M.SC., SIU-KI YU, PH.D., WAI-WANG LAM, M.SC., REBECCA WONG, PH.D., YICK-WING HO, M.SC., and SAU-FAN LIU, M.SC. Department of Radiotherapy, Hong Kong Sanatorium and Hospital, HongKong (Received 9 February 2010; accepted 24 June 2010)
Abstract—Tomotherapy adaptive dose calculation offers the ability to verify and adjust the therapeutic plan during the treatment. Using tomotherapy adaptive dose calculation, the planned fluence pattern can be used to recalculate the dose distribution on pretreatment megavoltage computed tomography (MVCT) images. Zipper artifacts, which appear as increased density in the central region of MVCT images, may affect the accuracy of adaptive dose recalculation. The purpose of this study was to evaluate the dosimetric effects of zipper artifacts on tomotherapy adaptive dose calculation. MVCT images of a cylindrical water phantom of 22-cm diameter were acquired on a tomotherapy system. The zipper artifacts were enclosed by a cylindrical planning target volume (PTV) contoured on these images. For comparison, artifact-free images were created by replacing the computed tomography (CT) numbers of zipper artifacts with the mean CT number of water. Treatment plans were generated by giving a uniform dose of 2 Gy to the PTV based on these modified images; it was then applied to the images that have the zipper artifacts. The impacts of different pitch ratios on the artifacts were assessed. The dose distribution differences between the 2 sets of images were compared. The absorbed dose that covered 95% volume of PTV and maximum dose, minimum dose, and mean dose of the PTV were also calculated and compared. The water phantom was scanned on the tomotherapy system twice per week for 12 consecutive weeks. The mean CT number of zipper artifacts (101 HU) was three times higher than that of water (34 HU). The CT number value and location of zipper artifacts were not affected by the pitch ratio. Gamma analysis was performed between the original and recalculated dose distributions. The discrepancies between the isodose distributions calculated by two sets of images were within 1%/1-mm tolerance. The dosimetric impact from zipper artifacts was found insignificant such that the recalculated dose was underestimated by less than 0.5%. © 2011 American Association of Medical Dosimetrists. Key words: TomoTherapy, Adaptive dose calculation, MVCT, Zipper artifact.
Tomotherapy uses the actual treatment beam as the x-ray source for image acquisition; no surrogate telemetry systems are required to register image space to treatment space. The MVCT scan contrast is linear, with respect to electron density of material imaged. This indicates that the MVCT images are appropriate for radiotherapy dose calculations in addition to image guidance of patient position. The accuracy and effectiveness of using MVCT images for dose recalculations has been reported.11–14 The imaging performance characteristics of the tomotherapy MVCT, which is characterized in terms of noise, uniformity, contrast, linearity, and spatial resolution, has been reported.15,16 These MVCT images could provide sufficient contrast to delineate many soft-tissue structures and perform dose recalculation. Beside these image characteristics, dose recalculation could be affected by MVCT artifacts. The central region of MVCT images, which were aligned with the tomotherapy central axis, showed artifacts with increased density. These artifacts that alternate from image to image on the reconstructed transverse slices will appear as a zipper artifact when viewed in sagittal or coronal plane, as shown in Fig. 1.
INTRODUCTION The imaging capability of a TomoTherapy Hi-Art II unit (TomoTherapy, Inc., Madison, WI) allows the acquisition of megavoltage computed tomography (MVCT) images of the patient in the treatment position before the delivery of rotational intensity-modulated radiation therapy.1,2 The purpose of these MVCT images is two-fold. First, the registration of these pretreatment MVCT images to the planning kilovoltage computed tomography (KVCT) images can be used to check and correct the patient’s position to ensure proper patient positioning for treatment delivery.3–7 Second, the daily MVCT images can be used to assess anatomical changes and evaluate the necessity of adjusting the therapeutic plan during the course of treatment to compensate for dosimetric errors that may have arisen from anatomical deformations.8 –10 This concept is referred to as adaptive radiotherapy. Presented at the 9th Asia Oceania Congress of Medical Physics and the 7th South East Asian Congress of Medical Physics on 23 October 2009, Chiang Mai, Thailand. Reprint requests to: Hui Geng, M. Sc., Department of Radiotherapy, G/F., Li Shu Pui Block, 2 Village Road, Happy Valley, Hong Kong. E-mail: [email protected] 306
Dosimetric effect of zipper artifacts on tomotherapy ● H. GENG et al.
307
Fig. 1. Zipper artifacts in brain.
The zipper artifacts arose from the isocenter misalignment.17 Same as conventional CT used in diagnostic radiography, the filtered back-projection method is used in the tomotherapy system for image reconstruction. The accuracy of the back-projection process depends on the precision of the alignment between the X-ray focal spot, the detector isochannel, and the center of rotation. In practice, however, it is unavoidable that a small amount of misalignment in the isocenter is present. For tomotherapy, the tolerance of X-ray source misalignment is 0.3 mm in both the IEC-x and IEC-y directions. Because the projection data are collected in helical mode and image reconstruction is based on half-scan data, the artifacts become periodical and appear as a zipper pattern in the coronal or sagittal view. The purpose of this study is to evaluate the dosimetric effects of zipper artifacts on tomotherapy adaptive dose calculation. METHOD AND MATERIALS The tomotherapy imaging system consists of a ring gantry with a xenon ion-chamber array mounted opposite the radiation source. The same linear accelerator is used for patient treatment and imaging. If used for imaging, the dose rate is reduced from 890 MU/min to 20 MU/ min. The radius of the detector is 110 cm, whereas the source-to-detector distance is 145 cm. This off-focus arrangement is used to improve the x-ray quantum detection efficiency for megavoltage beam. During MVCT acquisition, the beam is collimated to a length of 5 mm and a width of 400 mm at the isocenter plane. Images are acquired in helical mode. The only MVCT acquisition parameter that can be adjusted by the operator is the pitch ratio. This ratio is equal to the distance that the couch translates during one gantry rotation divided by the beam length at isocenter. The operator can choose one of three preselected pitch ratios that are labeled fine, normal, and coarse correspond to pitch ratios of 1, 1.6, and 2.4, respectively. The MVCT images are reconstructed by slice thicknesses of 2 mm, 4 mm, and 6 mm according to
pitch ratios of 1, 1.6, and 2.4. MVCT images are displayed as a 512 ⫻ 512 matrix with a 38.6-cm diameter field of view. A quality assurance water phantom (General Electric Medical Systems) as shown in Fig. 2 was used in this study. The phantom consists of a water-filled plastic cylinder in which a series of image quality test objects were positioned in the rear, with the front half containing water only. The diameter of the cylindrical phantom is 22 cm, whereas the front half, which contains water only, is 6 cm long. The phantom was immobilized with tape on the couch of the tomotherapy system. The axis of the cylindrical phantom was aligned with the rotational axis of the tomotherapy system, with the front half facing the gantry. MVCT images of the water phantom were acquired on the tomotherapy system twice per week for 12 consecutive weeks using different pitch ratios. The CT numbers of zipper artifact and water were monitored. One set of these MVCT images of the water phantom using pitch ratio of 1.6 was selected for planning. This pitch ratio was the most commonly used clinical
Fig. 2. Water phantom used in this study.
308
Medical Dosimetry
imaging setup in our hospital. A cylindrical planning target volume (PTV) was delineated on the MVCT images. The height and diameter of the PTV was 5 cm and 3 cm, respectively. During our previous investigations, it was found that zipper artifacts located within a circular region with a diameter of 3 cm were centered at the isocenter. Hence, the zipper artifacts were enclosed by the PTV. To evaluate the dosimetric impacts of zipper artifacts, an artifact-free MVCT image set was created by overriding the CT number of zipper artifacts to the mean CT number of water on the MVCT images using a commercial software (MATLAB, The MathWorks, Inc., Natick, MA), as shown in Fig. 3. The MVCT images of the water phantom were imported into the MATLAB program. The CT number of each pixel located within the PTV was replaced by a sum of two numbers. The first number is 34.2, which is the mean CT number of water (pitch ⫽ 1.6) monitored for 12 consecutive weeks. The second number is a random number generated by the MATLAB program. The value of this random number ranged between –27.1 and ⫹27.1 to simulate the standard deviation of the CT number of water. The artifact-free MVCT images were then transferred to the tomotherapy treatment planning system. Treatment plan was generated by prescribing a uniform dose of 2 Gy to the 95% volumes of PTV. Pretreatment MVCT images of the cylinder phantom were acquired on the tomotherapy system using all three pitch ratios. Three pretreatment MVCT images, which have the zipper artifacts, were registered to the artifact-free MVCT images, respectively, using PLANNED ADAPTIVE treatment planning software (TomoTherapy, Inc.). After registration, the beam characteristics of the original treatment plan generated from artifact-free MVCT images were applied to the pretreatment MVCT images and the radiation doses were recalculated.
Volume 36, Number 3, 2011
The dose distribution and dose volume histogram (DVH) of the PTV were compared between the original and the recalculated plans. To quantitatively compare the original and recalculated dose distributions, gamma analysis18 was performed using a 1% and 1-mm dose difference and distance-to-agreement criteria. The following dose end points were analyzed for the PTV: minimum, maximum, mean, and D95 (the mean dose irradiated to 95% volume of the PTV). The dose distribution on each slice was also reviewed and the under- or overdosed regions were identified. To minimize the influence of setup error, the water phantom was fixed at the same position during the whole image acquisition procedure. RESULTS Zipper artifacts were assessed for 12 consecutive weeks in all three pretreatment MVCT image sets, as shown in Fig. 4. The means and standard deviations of CT number for water were 33.6 ⫾ 26.3, 34.2 ⫾ 27.1, and 34.4 ⫾ 26.7, corresponding to 1, 1.6, and 2.4 pitch ratios, respectively. For zipper artifacts, the means and standard deviations of CT number were 101.0 ⫾ 28.3, 100.5 ⫾ 27.6, and 99.1 ⫾ 30.2 using pitch ratios of 1, 1.6, and 2.4, respectively. The CT number value and location of zipper artifacts were not affected by pitch ratio. The DVHs and dose distributions for the recalculated plans generated using three sets of pretreatment MVCT images were analyzed and compared with the original plan. A summary of the differences of the Dmin, Dmean, D95, and Dmax is shown in Table 1. None of the recalculated plans met the treatment prescription that delivers 2 Gy to 95% volume of the PTV. The Dmin, Dmean, and Dmax of the recalculated plans were lower than those of the original plan, but the dose differences were less than 0.5%.
Fig. 3. The CT number of zipper artifacts were override to the mean CT number of water. The red circles in the images were the PTV.
Dosimetric effect of zipper artifacts on tomotherapy ● H. GENG et al.
309
Fig. 4. Zipper artifacts acquired by 1, 1.6 and 2.4 pitch ratio, respectively.
In the comparison of the original and recalculated dose distributions, all calculated gamma indices are ⬍1, which indicates that the discrepancies between the original and recalculated dose planes were within 1%/1-mm tolerance. The isodose distributions on axial images of original and recalculated plans are shown in Fig. 5. The isodose lines were displayed on an absolute dose scale ranging from 1.96 –2.08 Gy. There was no significant difference on the 1.96-Gy (98%), 2.00-Gy (100%), and 2.04-Gy (102%) isodose lines between the recalculated and original plans. The volumes receiving doses higher than 2.08 Gy (104%) in the recalculated plans were less than those in the original plan. The cumulative DVHs of original and recalculated plans were plotted in Fig. 6. To emphasize the dose difference of different plans, the dose in the cumulative DVH was scaled from 1.98 –2.1 Gy. These dose differences were derived to differential DVHs as are shown in Fig. 7.
DISCUSSION Adaptive radiotherapy is a treatment technique that can systematically improve its treatment plan in response to patient temporal variations observed during the therapy process. Several techniques including kilovoltage cone-beam CT (KV CBCT)19 –21 and tomotherapy have been developed to correct the errors related to interfractional uncertainties of the treatment process. KV CBCT based on flat-panel technology integrated with a medical linear accelerator has become available from linear accelerator vendors for radiotherapy. KV CBCT typically consists of a kV source and flat-panel combination mounted on the drum of a medical accelerator, with the kV imaging axis orthogonal to that of the megavoltage therapy beam. KV CBCT has two important applications: patient setup and dose verification. Both applications rely on quality of the KV CBCT images. Compared with MVCT, KV CBCT im-
Table 1. A comparison of the Dmin, Dmean, Dmax, and D95 of the original plan and recalculated plans using three sets of pretreatment MVCT images
Original plan (artifact-free) Recalculated plan Pitch ⫽ 1 Recalculated plan Pitch ⫽ 1.6 Recalculated plan Pitch ⫽ 2.4
Dmin (Gy)
Dmean (Gy)
Dmax (Gy)
D95 (Gy)
1.983 1.976 (–0.35%) 1.981 (–0.10%) 1.983 (0.00%)
2.068 2.058 (–0.48%) 2.063 (–0.24%) 2.061 (–0.34%)
2.095 2.086 (–0.43%) 2.095 (0.00%) 2.090 (–0.24%)
2.000 1.994 (–0.30%) 1.999 (–0.05%) 1.997 (–0.15%)
310
Medical Dosimetry
Volume 36, Number 3, 2011
Fig. 5. Dose distribution comparisons of recalculated and original plans.
ages show better visualization of soft tissue. However, in the energy range used in KV CBCT, the photoelectric effect plays a significant role and the attenuation properties change with atomic number. As a result, metal artifacts appear in the KV CBCT images by the presence of high atomic number material such as dental fillings, surgical clips, and metal prostheses. Different from conventional CT, KV CBCT covers a much larger field of view (FOV) in the longitudinal direction. Hence, the KV CBCT image quality is affected by the increased amount of scattered photons caused by the large field size of x-rays. Corrections have to be made to use KV CBCT images for dose calculation. The feasibility of dose calculation using KV CBCT images has been investigated by many researchers.22–27 However, the gantry rotation speed is limited to about 1 minute, which makes the CBCT more prone to motion artifacts. When intrascanning organ motion is present, the dosimetric errors can be clinically significant, which limits the direct use of CBCT for dose calculation.28 Tomotherapy, on the other hand, adopted fan beam geometry to reduce the impact of scatter photons. Compton scattering effect is predominant in the MeV range used in MVCT scans, where attenuation coefficient is
almost independent of atomic number. The CT calibration curve is more linear in tomotherapy MVCT than in kV CBCT. Metal-induced artifacts can be minimized in MVCT images, as well. However, the gantry rotation period is 10 seconds for the TomoTherapy Hi-Art II unit. As a result, the qualities of tomotherapy MVCT images are also decreased by motion artifacts.29 To separate the dosimetric impacts on dose calculation from motion artifacts and zipper artifacts, a static phantom was used in this study. In this study, the dosimetric impacts on tomotherapy adaptive dose calculation resulting from zipper artifacts were evaluated for MVCT images acquired by different pitch ratios. Because the zipper artifacts arose from the isocenter misalignment, we expect the pitch should have little effect on the artifacts. The result presented in this study proved our expectance. As a result, recalculated dose was underestimated because of the increased attenuation coefficients introduced by zipper artifacts. The mean CT number of zipper artifacts is 101 HU, which correspond to an electron density of 1.067 calculated based on the tomotherapy MVCT calibration curve measured in our center. As a result, the electron density of zipper artifacts is equivalent to that of liver,
Fig. 6. Cumulative dose volume histograms of original and recalculated plans.
Dosimetric effect of zipper artifacts on tomotherapy ● H. GENG et al.
311
Fig. 7. Differential dose difference volume histograms of original and recalculated plans.
and the electron density uncertainty introduced by zipper artifacts results in no more than a 0.5% change in final dose calculation. Zipper artifacts were more pronounced in the rigid structures and organs, such as the brain. In the regions of the thorax and abdomen, the artifacts could be blurred by the motion of the organ or the heterogeneity of the structure. Patient positioning verification and correction were not affected by the zipper artifacts, because the artifacts were located within a circular region of diameter of 30 mm that centered at the isocenter, whereas the positioning correction was based on the patient’s skeleton which was not affected by the zipper artifacts. Although the dosimetric impacts of zipper artifacts on tomotherapy adaptive dose calculation were insignificant, the artifacts could decrease the MVCT image resolution and affect the accuracy of delineating anatomic structures, especially in the application of stereotactic radiotherapy. Care should be taken when delineating anatomical structures using MVCT images that contain zipper artifacts. CONCLUSION The dosimetric impact on tomotherapy adaptive dose calculation resulting from zipper artifacts was found to be clinically insignificant. REFERENCES 1. Yartsev, S.; Kron, T.; Dyk, J.V. Tomotherapy as a tool in imageguided radiation therapy: theoretical and technological aspects. Biomed. Imaging. Interv. J. 31:e16; 2007. 2. Mackie, T.R. History of tomotherapy. Phys. Med. Biol. 51:R427– R453; 2006. 3. Forrest, L.J.; Mackie, T.R.; Ruchala, K.; et al. The utility of megavoltage computed tomography images from a helical tomotherapy system for setup verification purposes. Int. J. Radiat. Oncol. Biol. Phys. 60:1639 – 44; 2004. 4. Sterzing, F. IGRT with helical tomotherapy-effort and benefit in clinical routine. Strahlenther. Onkol. 183:35–7; 2007.
5. Langen, K.M.; Zhang, Y.; Andrews, R.D.; et al. Initial experience with megavoltage (MV) CT guidance for daily prostate alignments. Int. J. Radiat. Oncol. Biol. Phys. 62:1517–24; 2005. 6. Chen, Y.J.; Han, C.; Liu, A.; et al. Setup variations in radiotherapy of esophageal cancer: evaluation by daily megavoltage computed tomographic localization. Int. J. Radiat. Oncol. Biol. Phys. 68: 1537– 45; 2007. 7. Boswell, S.A.; Jeraj, R.; Ruchala, K.; et al. A novel method to correct for pitch and yaw patient setup errors in helical tomotherapy. Med. Phys. 32:1630 –9; 2005. 8. Ramsey, C.R.; Langen, K.M.; Kupellian, P.A.; et al. A technique for adaptive image-guided helical tomotherapy for lung cancer. IntJ. Radiat. Oncol. Biol. Phys. 64:1237– 44; 2006. 9. Welsh, J.S.; Lock, M.; Harari, P.M.; et al. Clinical implementation of adaptive helical tomotherapy: a unique approach to imageguided intensity modulated radiotherapy. Technol. Cancer Res. Treat. 5:465–79; 2006. 10. Wu, C.; Jeraj, R.; Olivera, G.O.; et al. Re-optimization in adaptive radiotherapy. Phys. Med. Biol. 47:3181–95; 2002. 11. Ruchala, K.J.; Olivera, G.H.; Schloesser, E.A.; et al. Megavoltage CT on a tomotherapy system. Phys. Med. Biol. 44:2597– 621; 1999. 12. Ruchala, K.J.; Olivera, G.H.; Kapatoes, J.M.; et al. Megavoltage CT image reconstruction during tomotherapy treatments. Phys. Med. Biol. 45:3545– 62; 2000. 13. Ruchala, K.J.; Olivera, G.H.; Schloesser, E.A.; et al. Calibration of a tomotherapeutic MVCT system. Phys. Med. Biol. 45:N27–N36; 2000. 14. Ruchala, K.J.; Olivera, G.H.; Kapatoes, J.M.; et al. Megavoltage CT imaging as a by-product of multileaf collimator leakage. Phys. Med. Biol. 45:N61–N70; 2000. 15. Langen, K.M.; Meeks, S.L.; Poole, D.O.; et al. The use of megavoltage CT (MVCT) images for dose recomputations. Phys. Med. Biol. 50:4259 –76; 2005. 16. Meeks, S.L.; Harmon, J.F.; Langen, K.M.; et.al. Performance characterization of megavoltage computed tomography imaging on a helical tomotherapy unit. Med. Phys. 32:2673– 81; 2005. 17. Jiang, H. Three-dimensional artifact induced by projection weighting and misalignment. IEEE. Trans. Med. Imaging. 18:364 – 8; 1999. 18. Daniel, A.L.; William, B.H.; Sasa, M.; et al. A technique for the quantitative evaluation of dose distributions. Med. Phys. 25:656 – 61; 1998. 19. Jaffray, D.A.; Siewerdsen, J.H. Cone-beam computed tomography with a flat-panel imager: initial performance characterization. Med. Phys. 27:1311–23; 2000. 20. Jaffray, D.A.; Siewerdsen, J.H.; Wong, J.W.; et.al. Flat-panel cone-beam computed tomography for image-guided radiation therapy. IntJ. Radiat. Oncol. Biol. Phys. 53:1337– 49; 2002.
312
Medical Dosimetry
21. David, A.J.; Douglas, G.D.; Michel, M.; et al. A radiographic and tomographic imaging system integrated into a medical linear accelerator for localization of bone and soft-tissue targets. Int. J. Radiat. Oncol. Biol. Phys. 54:773– 89; 1999. 22. Uwe, O.; Thomas, T.; Simeon, N.; et al. Linac-integrated kV-cone beam CT: technical features and first applications. Med. Dosim. 31:62–70; 2006. 23. Sua,Y.; Fang-Fang, Y.. Dosimetric feasibility of cone-beam CTbased treatment planning compared to CT-based treatment planning. Int. J. Radiat. Oncol. Biol. Phys. 66:1553– 61; 2006. 24. George, X.D.; Dennis, M.D.; Charles, W.C.; et al. A study on adaptive IMRT treatment planning using KV cone-beam CT. Radiol. Oncol. 85:116 –25; 2007. 25. Louis, L.; Quynh-Thu, L.; Lei, X. Retrospective IMRT dose reconstruction based on cone-beam CT and MLC log-file. Int. J. Radiat. Oncol. Biol. Phys. 70:634 – 44; 2008.
Volume 36, Number 3, 2011 26. Joan, H.; Boyd, M.; Peter, B.G. Cone beam computerized tomography: the effect of calibration of the Hounsfield unit number to electron density on dose calculation accuracy for adaptive radiation therapy. Phys. Med. Biol. 54:N329 –N346; 2009. 27. Daniel, L.; Rebecca, W.; Douglas, M.; et al. Online planning and delivery technique for radiotherapy of spinal metastases using cone-beam CT: image quality and system performance. Int. J. Radiat. Oncol. Biol. Phys. 67:1229 –37; 2007. 28. Yong, Y.; Eduard, S.; Tianfang, L.; et al. Evaluation of on-board kV cone beam CT (CBCT)-based dose calculation. Phys. Med. Biol. 52:685–705; 2007. 29. Christopher, S.; Stewart, G.; Jerry, B. Delineation of moving targets with slow MVCT scans: implications for adaptive non-gated lung tomotherapy. Phys. Med. Biol. 52:1119 –34; 2007.
doi:10.1016/j.meddos.2010.06.002
THE DOSIMETRIC EFFECT OF ZIPPER ARTIFACTS ON TOMOTHERAPY ADAPTIVE DOSE CALCULATION—A PHANTOM STUDY WING-KEI
HUI GENG, M.SC., SIU-KI YU, PH.D., WAI-WANG LAM, M.SC., REBECCA WONG, PH.D., YICK-WING HO, M.SC., and SAU-FAN LIU, M.SC. Department of Radiotherapy, Hong Kong Sanatorium and Hospital, HongKong (Received 9 February 2010; accepted 24 June 2010)
Abstract—Tomotherapy adaptive dose calculation offers the ability to verify and adjust the therapeutic plan during the treatment. Using tomotherapy adaptive dose calculation, the planned fluence pattern can be used to recalculate the dose distribution on pretreatment megavoltage computed tomography (MVCT) images. Zipper artifacts, which appear as increased density in the central region of MVCT images, may affect the accuracy of adaptive dose recalculation. The purpose of this study was to evaluate the dosimetric effects of zipper artifacts on tomotherapy adaptive dose calculation. MVCT images of a cylindrical water phantom of 22-cm diameter were acquired on a tomotherapy system. The zipper artifacts were enclosed by a cylindrical planning target volume (PTV) contoured on these images. For comparison, artifact-free images were created by replacing the computed tomography (CT) numbers of zipper artifacts with the mean CT number of water. Treatment plans were generated by giving a uniform dose of 2 Gy to the PTV based on these modified images; it was then applied to the images that have the zipper artifacts. The impacts of different pitch ratios on the artifacts were assessed. The dose distribution differences between the 2 sets of images were compared. The absorbed dose that covered 95% volume of PTV and maximum dose, minimum dose, and mean dose of the PTV were also calculated and compared. The water phantom was scanned on the tomotherapy system twice per week for 12 consecutive weeks. The mean CT number of zipper artifacts (101 HU) was three times higher than that of water (34 HU). The CT number value and location of zipper artifacts were not affected by the pitch ratio. Gamma analysis was performed between the original and recalculated dose distributions. The discrepancies between the isodose distributions calculated by two sets of images were within 1%/1-mm tolerance. The dosimetric impact from zipper artifacts was found insignificant such that the recalculated dose was underestimated by less than 0.5%. © 2011 American Association of Medical Dosimetrists. Key words: TomoTherapy, Adaptive dose calculation, MVCT, Zipper artifact.
Tomotherapy uses the actual treatment beam as the x-ray source for image acquisition; no surrogate telemetry systems are required to register image space to treatment space. The MVCT scan contrast is linear, with respect to electron density of material imaged. This indicates that the MVCT images are appropriate for radiotherapy dose calculations in addition to image guidance of patient position. The accuracy and effectiveness of using MVCT images for dose recalculations has been reported.11–14 The imaging performance characteristics of the tomotherapy MVCT, which is characterized in terms of noise, uniformity, contrast, linearity, and spatial resolution, has been reported.15,16 These MVCT images could provide sufficient contrast to delineate many soft-tissue structures and perform dose recalculation. Beside these image characteristics, dose recalculation could be affected by MVCT artifacts. The central region of MVCT images, which were aligned with the tomotherapy central axis, showed artifacts with increased density. These artifacts that alternate from image to image on the reconstructed transverse slices will appear as a zipper artifact when viewed in sagittal or coronal plane, as shown in Fig. 1.
INTRODUCTION The imaging capability of a TomoTherapy Hi-Art II unit (TomoTherapy, Inc., Madison, WI) allows the acquisition of megavoltage computed tomography (MVCT) images of the patient in the treatment position before the delivery of rotational intensity-modulated radiation therapy.1,2 The purpose of these MVCT images is two-fold. First, the registration of these pretreatment MVCT images to the planning kilovoltage computed tomography (KVCT) images can be used to check and correct the patient’s position to ensure proper patient positioning for treatment delivery.3–7 Second, the daily MVCT images can be used to assess anatomical changes and evaluate the necessity of adjusting the therapeutic plan during the course of treatment to compensate for dosimetric errors that may have arisen from anatomical deformations.8 –10 This concept is referred to as adaptive radiotherapy. Presented at the 9th Asia Oceania Congress of Medical Physics and the 7th South East Asian Congress of Medical Physics on 23 October 2009, Chiang Mai, Thailand. Reprint requests to: Hui Geng, M. Sc., Department of Radiotherapy, G/F., Li Shu Pui Block, 2 Village Road, Happy Valley, Hong Kong. E-mail: [email protected] 306
Dosimetric effect of zipper artifacts on tomotherapy ● H. GENG et al.
307
Fig. 1. Zipper artifacts in brain.
The zipper artifacts arose from the isocenter misalignment.17 Same as conventional CT used in diagnostic radiography, the filtered back-projection method is used in the tomotherapy system for image reconstruction. The accuracy of the back-projection process depends on the precision of the alignment between the X-ray focal spot, the detector isochannel, and the center of rotation. In practice, however, it is unavoidable that a small amount of misalignment in the isocenter is present. For tomotherapy, the tolerance of X-ray source misalignment is 0.3 mm in both the IEC-x and IEC-y directions. Because the projection data are collected in helical mode and image reconstruction is based on half-scan data, the artifacts become periodical and appear as a zipper pattern in the coronal or sagittal view. The purpose of this study is to evaluate the dosimetric effects of zipper artifacts on tomotherapy adaptive dose calculation. METHOD AND MATERIALS The tomotherapy imaging system consists of a ring gantry with a xenon ion-chamber array mounted opposite the radiation source. The same linear accelerator is used for patient treatment and imaging. If used for imaging, the dose rate is reduced from 890 MU/min to 20 MU/ min. The radius of the detector is 110 cm, whereas the source-to-detector distance is 145 cm. This off-focus arrangement is used to improve the x-ray quantum detection efficiency for megavoltage beam. During MVCT acquisition, the beam is collimated to a length of 5 mm and a width of 400 mm at the isocenter plane. Images are acquired in helical mode. The only MVCT acquisition parameter that can be adjusted by the operator is the pitch ratio. This ratio is equal to the distance that the couch translates during one gantry rotation divided by the beam length at isocenter. The operator can choose one of three preselected pitch ratios that are labeled fine, normal, and coarse correspond to pitch ratios of 1, 1.6, and 2.4, respectively. The MVCT images are reconstructed by slice thicknesses of 2 mm, 4 mm, and 6 mm according to
pitch ratios of 1, 1.6, and 2.4. MVCT images are displayed as a 512 ⫻ 512 matrix with a 38.6-cm diameter field of view. A quality assurance water phantom (General Electric Medical Systems) as shown in Fig. 2 was used in this study. The phantom consists of a water-filled plastic cylinder in which a series of image quality test objects were positioned in the rear, with the front half containing water only. The diameter of the cylindrical phantom is 22 cm, whereas the front half, which contains water only, is 6 cm long. The phantom was immobilized with tape on the couch of the tomotherapy system. The axis of the cylindrical phantom was aligned with the rotational axis of the tomotherapy system, with the front half facing the gantry. MVCT images of the water phantom were acquired on the tomotherapy system twice per week for 12 consecutive weeks using different pitch ratios. The CT numbers of zipper artifact and water were monitored. One set of these MVCT images of the water phantom using pitch ratio of 1.6 was selected for planning. This pitch ratio was the most commonly used clinical
Fig. 2. Water phantom used in this study.
308
Medical Dosimetry
imaging setup in our hospital. A cylindrical planning target volume (PTV) was delineated on the MVCT images. The height and diameter of the PTV was 5 cm and 3 cm, respectively. During our previous investigations, it was found that zipper artifacts located within a circular region with a diameter of 3 cm were centered at the isocenter. Hence, the zipper artifacts were enclosed by the PTV. To evaluate the dosimetric impacts of zipper artifacts, an artifact-free MVCT image set was created by overriding the CT number of zipper artifacts to the mean CT number of water on the MVCT images using a commercial software (MATLAB, The MathWorks, Inc., Natick, MA), as shown in Fig. 3. The MVCT images of the water phantom were imported into the MATLAB program. The CT number of each pixel located within the PTV was replaced by a sum of two numbers. The first number is 34.2, which is the mean CT number of water (pitch ⫽ 1.6) monitored for 12 consecutive weeks. The second number is a random number generated by the MATLAB program. The value of this random number ranged between –27.1 and ⫹27.1 to simulate the standard deviation of the CT number of water. The artifact-free MVCT images were then transferred to the tomotherapy treatment planning system. Treatment plan was generated by prescribing a uniform dose of 2 Gy to the 95% volumes of PTV. Pretreatment MVCT images of the cylinder phantom were acquired on the tomotherapy system using all three pitch ratios. Three pretreatment MVCT images, which have the zipper artifacts, were registered to the artifact-free MVCT images, respectively, using PLANNED ADAPTIVE treatment planning software (TomoTherapy, Inc.). After registration, the beam characteristics of the original treatment plan generated from artifact-free MVCT images were applied to the pretreatment MVCT images and the radiation doses were recalculated.
Volume 36, Number 3, 2011
The dose distribution and dose volume histogram (DVH) of the PTV were compared between the original and the recalculated plans. To quantitatively compare the original and recalculated dose distributions, gamma analysis18 was performed using a 1% and 1-mm dose difference and distance-to-agreement criteria. The following dose end points were analyzed for the PTV: minimum, maximum, mean, and D95 (the mean dose irradiated to 95% volume of the PTV). The dose distribution on each slice was also reviewed and the under- or overdosed regions were identified. To minimize the influence of setup error, the water phantom was fixed at the same position during the whole image acquisition procedure. RESULTS Zipper artifacts were assessed for 12 consecutive weeks in all three pretreatment MVCT image sets, as shown in Fig. 4. The means and standard deviations of CT number for water were 33.6 ⫾ 26.3, 34.2 ⫾ 27.1, and 34.4 ⫾ 26.7, corresponding to 1, 1.6, and 2.4 pitch ratios, respectively. For zipper artifacts, the means and standard deviations of CT number were 101.0 ⫾ 28.3, 100.5 ⫾ 27.6, and 99.1 ⫾ 30.2 using pitch ratios of 1, 1.6, and 2.4, respectively. The CT number value and location of zipper artifacts were not affected by pitch ratio. The DVHs and dose distributions for the recalculated plans generated using three sets of pretreatment MVCT images were analyzed and compared with the original plan. A summary of the differences of the Dmin, Dmean, D95, and Dmax is shown in Table 1. None of the recalculated plans met the treatment prescription that delivers 2 Gy to 95% volume of the PTV. The Dmin, Dmean, and Dmax of the recalculated plans were lower than those of the original plan, but the dose differences were less than 0.5%.
Fig. 3. The CT number of zipper artifacts were override to the mean CT number of water. The red circles in the images were the PTV.
Dosimetric effect of zipper artifacts on tomotherapy ● H. GENG et al.
309
Fig. 4. Zipper artifacts acquired by 1, 1.6 and 2.4 pitch ratio, respectively.
In the comparison of the original and recalculated dose distributions, all calculated gamma indices are ⬍1, which indicates that the discrepancies between the original and recalculated dose planes were within 1%/1-mm tolerance. The isodose distributions on axial images of original and recalculated plans are shown in Fig. 5. The isodose lines were displayed on an absolute dose scale ranging from 1.96 –2.08 Gy. There was no significant difference on the 1.96-Gy (98%), 2.00-Gy (100%), and 2.04-Gy (102%) isodose lines between the recalculated and original plans. The volumes receiving doses higher than 2.08 Gy (104%) in the recalculated plans were less than those in the original plan. The cumulative DVHs of original and recalculated plans were plotted in Fig. 6. To emphasize the dose difference of different plans, the dose in the cumulative DVH was scaled from 1.98 –2.1 Gy. These dose differences were derived to differential DVHs as are shown in Fig. 7.
DISCUSSION Adaptive radiotherapy is a treatment technique that can systematically improve its treatment plan in response to patient temporal variations observed during the therapy process. Several techniques including kilovoltage cone-beam CT (KV CBCT)19 –21 and tomotherapy have been developed to correct the errors related to interfractional uncertainties of the treatment process. KV CBCT based on flat-panel technology integrated with a medical linear accelerator has become available from linear accelerator vendors for radiotherapy. KV CBCT typically consists of a kV source and flat-panel combination mounted on the drum of a medical accelerator, with the kV imaging axis orthogonal to that of the megavoltage therapy beam. KV CBCT has two important applications: patient setup and dose verification. Both applications rely on quality of the KV CBCT images. Compared with MVCT, KV CBCT im-
Table 1. A comparison of the Dmin, Dmean, Dmax, and D95 of the original plan and recalculated plans using three sets of pretreatment MVCT images
Original plan (artifact-free) Recalculated plan Pitch ⫽ 1 Recalculated plan Pitch ⫽ 1.6 Recalculated plan Pitch ⫽ 2.4
Dmin (Gy)
Dmean (Gy)
Dmax (Gy)
D95 (Gy)
1.983 1.976 (–0.35%) 1.981 (–0.10%) 1.983 (0.00%)
2.068 2.058 (–0.48%) 2.063 (–0.24%) 2.061 (–0.34%)
2.095 2.086 (–0.43%) 2.095 (0.00%) 2.090 (–0.24%)
2.000 1.994 (–0.30%) 1.999 (–0.05%) 1.997 (–0.15%)
310
Medical Dosimetry
Volume 36, Number 3, 2011
Fig. 5. Dose distribution comparisons of recalculated and original plans.
ages show better visualization of soft tissue. However, in the energy range used in KV CBCT, the photoelectric effect plays a significant role and the attenuation properties change with atomic number. As a result, metal artifacts appear in the KV CBCT images by the presence of high atomic number material such as dental fillings, surgical clips, and metal prostheses. Different from conventional CT, KV CBCT covers a much larger field of view (FOV) in the longitudinal direction. Hence, the KV CBCT image quality is affected by the increased amount of scattered photons caused by the large field size of x-rays. Corrections have to be made to use KV CBCT images for dose calculation. The feasibility of dose calculation using KV CBCT images has been investigated by many researchers.22–27 However, the gantry rotation speed is limited to about 1 minute, which makes the CBCT more prone to motion artifacts. When intrascanning organ motion is present, the dosimetric errors can be clinically significant, which limits the direct use of CBCT for dose calculation.28 Tomotherapy, on the other hand, adopted fan beam geometry to reduce the impact of scatter photons. Compton scattering effect is predominant in the MeV range used in MVCT scans, where attenuation coefficient is
almost independent of atomic number. The CT calibration curve is more linear in tomotherapy MVCT than in kV CBCT. Metal-induced artifacts can be minimized in MVCT images, as well. However, the gantry rotation period is 10 seconds for the TomoTherapy Hi-Art II unit. As a result, the qualities of tomotherapy MVCT images are also decreased by motion artifacts.29 To separate the dosimetric impacts on dose calculation from motion artifacts and zipper artifacts, a static phantom was used in this study. In this study, the dosimetric impacts on tomotherapy adaptive dose calculation resulting from zipper artifacts were evaluated for MVCT images acquired by different pitch ratios. Because the zipper artifacts arose from the isocenter misalignment, we expect the pitch should have little effect on the artifacts. The result presented in this study proved our expectance. As a result, recalculated dose was underestimated because of the increased attenuation coefficients introduced by zipper artifacts. The mean CT number of zipper artifacts is 101 HU, which correspond to an electron density of 1.067 calculated based on the tomotherapy MVCT calibration curve measured in our center. As a result, the electron density of zipper artifacts is equivalent to that of liver,
Fig. 6. Cumulative dose volume histograms of original and recalculated plans.
Dosimetric effect of zipper artifacts on tomotherapy ● H. GENG et al.
311
Fig. 7. Differential dose difference volume histograms of original and recalculated plans.
and the electron density uncertainty introduced by zipper artifacts results in no more than a 0.5% change in final dose calculation. Zipper artifacts were more pronounced in the rigid structures and organs, such as the brain. In the regions of the thorax and abdomen, the artifacts could be blurred by the motion of the organ or the heterogeneity of the structure. Patient positioning verification and correction were not affected by the zipper artifacts, because the artifacts were located within a circular region of diameter of 30 mm that centered at the isocenter, whereas the positioning correction was based on the patient’s skeleton which was not affected by the zipper artifacts. Although the dosimetric impacts of zipper artifacts on tomotherapy adaptive dose calculation were insignificant, the artifacts could decrease the MVCT image resolution and affect the accuracy of delineating anatomic structures, especially in the application of stereotactic radiotherapy. Care should be taken when delineating anatomical structures using MVCT images that contain zipper artifacts. CONCLUSION The dosimetric impact on tomotherapy adaptive dose calculation resulting from zipper artifacts was found to be clinically insignificant. REFERENCES 1. Yartsev, S.; Kron, T.; Dyk, J.V. Tomotherapy as a tool in imageguided radiation therapy: theoretical and technological aspects. Biomed. Imaging. Interv. J. 31:e16; 2007. 2. Mackie, T.R. History of tomotherapy. Phys. Med. Biol. 51:R427– R453; 2006. 3. Forrest, L.J.; Mackie, T.R.; Ruchala, K.; et al. The utility of megavoltage computed tomography images from a helical tomotherapy system for setup verification purposes. Int. J. Radiat. Oncol. Biol. Phys. 60:1639 – 44; 2004. 4. Sterzing, F. IGRT with helical tomotherapy-effort and benefit in clinical routine. Strahlenther. Onkol. 183:35–7; 2007.
5. Langen, K.M.; Zhang, Y.; Andrews, R.D.; et al. Initial experience with megavoltage (MV) CT guidance for daily prostate alignments. Int. J. Radiat. Oncol. Biol. Phys. 62:1517–24; 2005. 6. Chen, Y.J.; Han, C.; Liu, A.; et al. Setup variations in radiotherapy of esophageal cancer: evaluation by daily megavoltage computed tomographic localization. Int. J. Radiat. Oncol. Biol. Phys. 68: 1537– 45; 2007. 7. Boswell, S.A.; Jeraj, R.; Ruchala, K.; et al. A novel method to correct for pitch and yaw patient setup errors in helical tomotherapy. Med. Phys. 32:1630 –9; 2005. 8. Ramsey, C.R.; Langen, K.M.; Kupellian, P.A.; et al. A technique for adaptive image-guided helical tomotherapy for lung cancer. IntJ. Radiat. Oncol. Biol. Phys. 64:1237– 44; 2006. 9. Welsh, J.S.; Lock, M.; Harari, P.M.; et al. Clinical implementation of adaptive helical tomotherapy: a unique approach to imageguided intensity modulated radiotherapy. Technol. Cancer Res. Treat. 5:465–79; 2006. 10. Wu, C.; Jeraj, R.; Olivera, G.O.; et al. Re-optimization in adaptive radiotherapy. Phys. Med. Biol. 47:3181–95; 2002. 11. Ruchala, K.J.; Olivera, G.H.; Schloesser, E.A.; et al. Megavoltage CT on a tomotherapy system. Phys. Med. Biol. 44:2597– 621; 1999. 12. Ruchala, K.J.; Olivera, G.H.; Kapatoes, J.M.; et al. Megavoltage CT image reconstruction during tomotherapy treatments. Phys. Med. Biol. 45:3545– 62; 2000. 13. Ruchala, K.J.; Olivera, G.H.; Schloesser, E.A.; et al. Calibration of a tomotherapeutic MVCT system. Phys. Med. Biol. 45:N27–N36; 2000. 14. Ruchala, K.J.; Olivera, G.H.; Kapatoes, J.M.; et al. Megavoltage CT imaging as a by-product of multileaf collimator leakage. Phys. Med. Biol. 45:N61–N70; 2000. 15. Langen, K.M.; Meeks, S.L.; Poole, D.O.; et al. The use of megavoltage CT (MVCT) images for dose recomputations. Phys. Med. Biol. 50:4259 –76; 2005. 16. Meeks, S.L.; Harmon, J.F.; Langen, K.M.; et.al. Performance characterization of megavoltage computed tomography imaging on a helical tomotherapy unit. Med. Phys. 32:2673– 81; 2005. 17. Jiang, H. Three-dimensional artifact induced by projection weighting and misalignment. IEEE. Trans. Med. Imaging. 18:364 – 8; 1999. 18. Daniel, A.L.; William, B.H.; Sasa, M.; et al. A technique for the quantitative evaluation of dose distributions. Med. Phys. 25:656 – 61; 1998. 19. Jaffray, D.A.; Siewerdsen, J.H. Cone-beam computed tomography with a flat-panel imager: initial performance characterization. Med. Phys. 27:1311–23; 2000. 20. Jaffray, D.A.; Siewerdsen, J.H.; Wong, J.W.; et.al. Flat-panel cone-beam computed tomography for image-guided radiation therapy. IntJ. Radiat. Oncol. Biol. Phys. 53:1337– 49; 2002.
312
Medical Dosimetry
21. David, A.J.; Douglas, G.D.; Michel, M.; et al. A radiographic and tomographic imaging system integrated into a medical linear accelerator for localization of bone and soft-tissue targets. Int. J. Radiat. Oncol. Biol. Phys. 54:773– 89; 1999. 22. Uwe, O.; Thomas, T.; Simeon, N.; et al. Linac-integrated kV-cone beam CT: technical features and first applications. Med. Dosim. 31:62–70; 2006. 23. Sua,Y.; Fang-Fang, Y.. Dosimetric feasibility of cone-beam CTbased treatment planning compared to CT-based treatment planning. Int. J. Radiat. Oncol. Biol. Phys. 66:1553– 61; 2006. 24. George, X.D.; Dennis, M.D.; Charles, W.C.; et al. A study on adaptive IMRT treatment planning using KV cone-beam CT. Radiol. Oncol. 85:116 –25; 2007. 25. Louis, L.; Quynh-Thu, L.; Lei, X. Retrospective IMRT dose reconstruction based on cone-beam CT and MLC log-file. Int. J. Radiat. Oncol. Biol. Phys. 70:634 – 44; 2008.
Volume 36, Number 3, 2011 26. Joan, H.; Boyd, M.; Peter, B.G. Cone beam computerized tomography: the effect of calibration of the Hounsfield unit number to electron density on dose calculation accuracy for adaptive radiation therapy. Phys. Med. Biol. 54:N329 –N346; 2009. 27. Daniel, L.; Rebecca, W.; Douglas, M.; et al. Online planning and delivery technique for radiotherapy of spinal metastases using cone-beam CT: image quality and system performance. Int. J. Radiat. Oncol. Biol. Phys. 67:1229 –37; 2007. 28. Yong, Y.; Eduard, S.; Tianfang, L.; et al. Evaluation of on-board kV cone beam CT (CBCT)-based dose calculation. Phys. Med. Biol. 52:685–705; 2007. 29. Christopher, S.; Stewart, G.; Jerry, B. Delineation of moving targets with slow MVCT scans: implications for adaptive non-gated lung tomotherapy. Phys. Med. Biol. 52:1119 –34; 2007.