- Email: [email protected]
Analysis of ion beam teletherapy patient-specific quality assurance
ARTICLE IN PRESS Medical Dosimetry ■■ (2018) ■■–■■
Medical Dosimetry j o u r n a l h o m e p a g e : w w w. m e d d o s . o r g
Medical Physics Contribution:
Analysis of ion beam teletherapy patient-specific quality assurance Xiaoli Liu, M.S., Yu Deng, M.S., Nicki Schlegel, M.S., Zhijie Huang, Ph.D., and Michael F. Moyers, Ph.D. Department of Medical Physics, Shanghai Proton and Heavy Ion Center, Shanghai 201315, China
A R T I C L E
I N F O
Article history:
Received 25 April 2017 Accepted 25 January 2018 Keywords:
Quality assurance Ion beam Modulated scanning
A B S T R A C T
The objective of this study was to evaluate the procedures for patient-specific quality assurance measurements using modulated scanned and energy stacked beams for proton and carbon ion teletherapy. Delivery records from 1734 portal measurements were analyzed using a 3-point pass criteria: more than 22 of 24 chambers in a water phantom (WP) had to have a measured dose difference from the planned portal doses less than or equal to 3%, or the distance from the measurement point location to a point location in the plan having the same dose had to be less than or equal to 3 mm (distance to agreement [DTA]), and the mean dose deviation of all chambers had to be less than 3%. Stratification of results showed some associations between measurement parameters and pass rates. For proton portals, pass rates were high at all measurement depths, but for carbon ion portals, pass rates decreased as a function of increasing measurement depth. Pass rates of both proton and carbon ion portals with 1 WP were slightly lower than those with a second WP. The total pass rates were 97.7% and 91.9% for proton and carbon ion patient portals, respectively. In general, the measured doses exhibited good agreement with the treatment planning system (TPS) calculated doses. When the chamber position was deeper than 150 mm in carbon ion beams, a lower pass rate was observed, which may have been caused by ion chamber array setup uncertainty (lateral and depth) in highly modulated portals or incorrect modeling of scatter by the TPS. These deviations need further investigation. © 2018 American Association of Medical Dosimetrists.
Introduction Teletherapy beams of light ions, such as protons and carbons, share the physical advantage of a finite range in depth with a steep distal dose gradient after the so-called Bragg peak.1-3 By combining multiple energies to place the
Reprint requests to Xiaoli Liu, M.D., Department of Medical Physics, Shanghai Proton and Heavy Ion Center, Shanghai 201315, China. E-mail: [email protected]
Bragg peak at different depths throughout the target, a conformal dose distribution can be delivered to the target while minimizing radiation effects in the surrounding normal tissues. In the year 2014, the Shanghai Proton and Heavy Ion Center (SPHIC) opened and performed a successful clinical trial of patients using a Siemens Iontris treatment delivery system.4 This system provides modulated scanned and energy stacked proton and carbon ion beams with ranges in water up to 310 mm (energy per nucleons of 221 MeV/n for proton and 430 MeV/n for carbon ions)5-7; this system does not currently support the use of apertures or boluses. The SPHIC facility has 4 ion beam treatment rooms, 3 rooms with
https://doi.org/10.1016/j.meddos.2018.01.002 0958-3947/Copyright © 2018 American Association of Medical Dosimetrists
ARTICLE IN PRESS 2
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
horizontally oriented beamlines, and 1 room with a beamline at a 45° incline. All rooms are equipped with robotic patient positioners with an exchangeable flat table top mounted onto a robotic arm. Each treatment room also has an imaging robot mounted on the ceiling to verify patient alignment before treatment. Routine patient operations began after the clinical trial. Similar to the introduction of dynamic multileaf collimators into radiotherapy practice using megavoltage × ray beams, the move to modulated scanned and energy stacked ion beams also requires additional and efficient methods of patient-specific quality assurance (QA).8-10 Between 2015 and 2016, patient-specific QA was performed using a clustered array of ion chambers in a water phantom (WP). This report provides a summary and analysis of 1734 portal delivery records during the period from July 2015 to July 2016 during routine operations.
Fig. 1. Water phantom setup at isocenter for measurement of patientspecific portal. The inset at the upper right is a close-up picture of the ionization chamber array. (Color version of figure is available online.)
Methods and Materials At SPHIC, a QA procedure was performed for every patient portal before it was delivered to a patient. The current QA procedure was similar to that which was performed at the Gesellschaft für Schwerionenforschung and the Heidelberg Ion Therapy Center.11,12 This procedure consisted of copying all the portal parameters (spot positions, energies, numbers of ions, use of a beamline range shifter [RS], etc.) from each portal of a patient plan into a so-called verification plan that consisted of a rectangular parallelepiped WP, with each portal being directed into the side of the phantom. A verification plan may include more than 1 portal used for a given patient during a given treatment fraction, but in the verification plan, all portals entered the phantom from the same direction. The dose distribution of each portal was then recalculated, the verification plan was transmitted to the beam delivery system, and then all portals of this verification plan were delivered with an arbitrary horizontal beamline room. The use of any room for the verification measurement was possible because the beam range and beam monitors in all rooms were calibrated identically to within the clinical tolerance. The ion beam treatment planning system (TPS) used at SPHIC was Syngo RT Planning VC11B (Siemens, Erlangen, Germany). The WP input into the TPS for verification planning represented a real WP (PTW, Freiburg, Germany) that contained an array of 24 miniature thimble chambers in a polymethylmethacrylate block attached to a 3-dimensional scanning arm.13 The 24 chambers (PTW model 31015, PTW) each had an active volume of 0.03 cm3 and were arranged in 6 rows perpendicular to the beam axis (i.e., 6 depths), with each row containing 4 chambers. Every other row was offset laterally from the one in front, and every second row was offset slightly in the vertical direction to reduce perturbations
of the delivered dose distributions at chamber locations at deeper depths (see Fig. 1 inset). The chamber array covered a cubical volume of approximately 55 mm × 70 mm × 60 mm, and the planner could place this volume at selected locations within the WP to measure the absolute dose both in the center of the target and its periphery. In addition to providing the expected dose at each chamber location of the array, the TPS also provided the dose gradient at each chamber location to assist in the analysis of potential dose deviations. The dose calculations did not include the effects of the ion chamber air cavities or the effects of the polymethylmethacrylate block. Figure 1 shows the phantom setup in the treatment room. The facility utilized 2 separate but identical WPs and chamber arrays to measure patient portals, WP 1 and WP 2. Typically, only 1 phantom was used on a given day with the other one being used only for backup in case the primary one was not working properly or the chambers and the electrometer associated with the phantom had been sent for calibration. Because the WP entrance surface was always perpendicular to the floor, it could not be used with the oblique beamline (room 3); therefore, all measurements are performed in one of the horizontal rooms (rooms 1, 2, and 4). When routine clinical operations started, measurements were made at 3 different positions of the chamber array block for every portal, typically near the surface, in the middle of the target, and near the distal end of the target. After several months of operation, it was discovered that the pass rates for chamber locations near the surface were usually very high because the dose gradients in this region were typically quite small; thereafter, only 1 array block position was typically used and chosen to be located within the high-dose target volume. If the target was large, 2 positions were sometimes chosen, 1 in the middle of the target and 1 near the
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
3
Fig. 2. TPS display of a single portal of a verification plan. The upper left panel represents a view of the dose distribution in a vertical plane as seen from the side of the phantom. The upper right panel represents a view of the dose distribution in a horizontal plane as seen from the top of the phantom. The lower left panel represents a view of the dose distribution in a vertical plane perpendicular to the beam axis. The lower right panel is similar to the upper right panel but shows each ion chamber as a circle and a display of the absolute dose per fraction and dose gradient for each chamber. The color for each circle matches the isodose color levels. (Color version of figure is available online.)
distal end of the target to check the beam range. A TPS display of a single portal from a verification plan is shown in Fig. 2. The scanning WP was connected to the beam delivery system via an Ethernet connection to enable automatic setup of the array positions according to the verification plan. During measurement, the delivery system was run in a medical delivery mode with all beam delivery interlocks and tolerances identical to those used during treatment. To check the integrity of the WP acquisition system setup before each
measurement session of patient verification plans, standard plans for cubical targets were measured. Simultaneously, the dose per MU for the standard plans was checked. The standard plans for both proton and carbon ion beams were designed for cubical targets of 80 mm × 80 mm × 80 mm and a physical dose of 2 Gy per fraction. The depths for the center of the front row of chambers for the chamber array for the 5 standard proton cube plans were 36.3, 70.2, 125.2, 175.2, and 222.2 mm. The depths for the center of the front row of chambers for the chamber array for the 5 standard carbon
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
4
Table 1 Stratification of measurements
Carbon cubes Proton cubes Carbon portals Proton portals
Total
Water phantom 1
Water phantom 2
Depth 0-50 mm
Depth 50-100 mm
Depth 100-150 mm
Depth 150-200 mm
Room 1
Room 2
Room 4
136 109 1211 523
30 70 296 191
106 39 916 332
23 15 790 361
7 53 278 139
99 39 131 22
7 2 12 1
45 34 383 122
41 40 365 219
50 35 463 182
ion cube plans were 36.4, 71.3, 121.3, 171.3, and 221.3 mm. As seen in Table 1, cube plan 2 was most often used for proton measurement sessions and cube plan 3 was most often used for carbon ion measurement sessions. Patient-specific verification measurements were done in horizontal rooms using either WP 1 or WP 2. After portal delivery, the beam delivery system saved all the delivery and measurement records in an XML formatted file, including all beam parameters, the depth of the first row of chambers, and the beamline identification. The file was then imported into Excel for reading the information and performing analysis. From July 2015 to July 2016, a total of 1734 portal measurements (523 protons and 1211 carbon ions) from 608 plans were performed at SPHIC with some of these being repeat measurements.
Results and Discussion Table 1 shows the number of measurements for cube and patient verification plans stratified by WP, depth of
ion chamber block in phantom, and measurement room. Cubical target tests During the period of the present study, there were 245 cube measurements. Figure 3 shows the dose deviations of the cube measurements from their expected values stratified by the different categories described in the Methods and Materials section. In general, the measured cube doses matched the expected plan doses to better than ± 1%. The proton cube measurements showed smaller deviations than the carbon ion cube measurements. To investigate the behavior of each chamber of the array, the mean percentage dose deviations of each chamber for proton and carbon ion cube plans are plotted in Fig. 4. The mean percentage deviation for all chambers was within ± 2% with chambers 4 and 5 of WP 1 measuring the lowest doses for proton portals (− 1.56 ± 0.93% for chamber 4 and − 1.04 ± 0.74% for chamber 5). A similar phenomenon was found for proton measurements using WP 2.
Fig. 3. Mean percentage dose deviations for proton and carbon ion cube measurements stratified by WP, depth of ion chamber block in phantom, and measurement room. Plotted error bars represent 1 standard deviation from the mean. (Color version of figure is available online.)
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
5
Fig. 4. Mean percentage dose deviation for each chamber for proton and carbon ion cube measurements in water phantoms 1 and 2. Plotted error bars represent ± 1 standard deviation of each chamber. (Color version of figure is available online.)
Patient verification measurements The current criteria for a chamber measurement to pass are either that the deviation of the measured dose from the predicted planned dose is less than or equal to 3% or the distance between the location of a measured data point and the nearest point in the planned dose distribution that has the same dose is less than or equal to 3 mm (referred to as distance to agreement [DTA]). The portal verification is considered to pass if more than 22 of the 24 chambers pass these criteria and the mean dose deviation of all the chambers is less than or equal to 3%. During the study period, 523 proton portal measurements were performed. The absolute value of the mean percentage dose deviation between the measured and
planned doses for all chambers of the array was typically less than 1% for each measurement. For portals that included a RS, the mean of the mean dose deviation for all chambers of the array increased to − 1.19 ± 1.11%. Figure 5 shows the percentage mean dose deviations for patient verification measurements for proton portals stratified by WP, room, and chamber array depth. Figure 6 shows the percentage mean dose deviations for all proton portals as a function of ionization chamber. The mean deviation from the plan doses for most chambers was within ± 1%. The two chambers with the largest mean deviations were chamber 20 of WP 2 (+ 1.18 ± 3.05%) and chamber 5 of WP 1 (− 1.02 ± 2.13%). During the study period, 1211 carbon ion portal measurements were performed. Figure 7 shows the percentage
Fig. 5. Percentage mean dose deviation for proton portals stratified by WP, RS, depth of ion chamber array in phantom, and measurement room. Plotted error bars represent 1 standard deviation of the mean. (Color version of figure is available online.)
ARTICLE IN PRESS 6
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
Fig. 6. Percentage mean dose deviations of each chamber for all proton portals. Error bars represent 1 standard deviation of each chamber using water phantoms 1 and 2. (Color version of figure is available online.)
mean dose deviations for carbon ion portals stratified by WP, room, and chamber array depth. The absolute value of the percentage mean deviation between the measured and planned doses for all stratifications was less than 2.0%. For portals that included a RS, the percentage mean deviation was − 1.95 ± 1.46%. Figure 8 shows the percentage mean dose deviations for carbon ion portals as a function of ionization chamber. The absolute value of the percentage mean deviation for most chambers was within 2%. The 2 chambers with the largest mean deviations were chamber 5 of WP 1 (− 2.47 ± 2.30%) and chamber 20 of WP 2 (+ 0.09 ± 3.58%).
As seen in Figs. 6 and 8, the pass rates for proton and carbon ion ports were higher when using WP 2 than when using WP 1. This finding may be a result of the ion chamber calibration factors for phantom 2 being slightly higher than those for WP 1. The results also show that, on average, both proton and carbon ion portal measured doses were lower than the predicted planned doses. The total pass rates for proton and carbon ion portals were 97.7% and 91.9%, respectively. The percentage pass rates for all patient portals stratified by the parameters described previously are plotted in Fig. 9. The pass rate for carbon ion portals was seen to be a function of detector array depth,
Fig. 7. Percentage mean dose deviations for carbon ion portals stratified by WP, RS, depth of ion chamber array in phantom, and measurement room. Error bars represent 1 standard deviation of the mean. (Color version of figure is available online.)
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
7
Fig. 8. Percentage mean dose deviations of each chamber for carbon ion portals. Plotted error bars represent 1 standard deviation of each chamber using water phantoms 1 and 2. (Color version of figure is available online.)
with the lowest pass rate being observed when the chamber array depth was between 150 and 200 mm (66.7%). It should be noted, however, that this deepest stratification category contained only a few measurements (n = 13). This phenomenon needs further investigation. For proton portals, no significant correlation with depth was observed. The total pass rates for each of the individual chamber array positions in both WPs for all proton and carbon ion portal measurements are plotted in Fig. 10. The lowest pass rates for any chamber position were found in WP 1 for chambers 4 and 5 (89.2% and 83.4%, respectively) when used with the carbon ion beam. This phantom and chamber array was used infrequently but, in May 2016, a comprehensive check of the system revealed a slightly higher leakage rate for these 2 chamber channels. For portals with long delivery times
(>5 minutes), this finding may have resulted in a 1.6% to 2.0% lower reported dose, causing the chamber measurements to fail. There have been insufficient data taken after repairs by the manufacturer to evaluate if this leakage was indeed the cause of the lower pass rate. The total pass rates of portals from both patient-specific and standard cube plans as a function of time are plotted in Fig. 11. Before November 2015, WP 1 was the primary measurement device for verification plans. In November of 2015, the chambers for this phantom were sent for calibration and WP 2 was then used as the primary measurement device. Subsequent to this change, the pass rate for patientspecific portals was seen to increase. Out of a total of 1734 portal measurements from 608 plans (including some remeasured plans), 56 plans failed to pass
Fig. 9. Proton and carbon ion patient-specific portal pass rate stratified by WP, RS, and depth of ion chamber array in phantom. (Color version of figure is available online.)
ARTICLE IN PRESS 8
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
Fig. 10. Pass rates of each chamber position of the array for proton and carbon ion patient-specific portal measurements. (Color version of figure is available online.)
the criteria of 22 of 24 chambers having less than or equal to 3% dose deviations, 22 of 24 chambers having a DTA of less than or equal to 3 mm, and a mean dose deviation for all chambers less than 3%. A verification plan is considered to fail if any of the portals of that plan failed. Some plans that originally failed passed by remeasuring the failed portal or by shifting the chamber array to account for slight setup uncertainties. Forty-nine of the failed plans occurred for carbon ion plans and 7 for proton plans. Each of those failures required a more critical analysis. Table 2 summarizes the major plan parameters for those failures. Most of the failures were caused by large dose deviations of individual chambers. If the pass criteria were set to use a gamma index14 with a 3% dose difference or 3-mm DTA instead of the 3-point
criteria evaluation described above, then only 6 of the 608 plan measurements would have been considered to have failed. Four of those failed cases were for prostate treatments using carbon ion beams. These failed because of the mean dose deviation being larger than 3%. In these cases, the measurements were reperformed with the chamber array at slightly different locations, resulting in a pass for each case. One of the failed cases was for an adenoid cystic carcinoma treated with a carbon ion beam and a RS. This case also failed because of the mean dose deviation being larger than 3%. In this case, the plan was reoptimized, after which the verification measurement passed. Only 1 proton beam case was found to fail using the individual criteria test. Upon investigation, it was found that the prescription was set to the
Fig. 11. Percentage pass rate for both standard cube plans and patient-specific portal measurements as a function of time binned into monthly intervals. (Color version of figure is available online.)
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
9
Table 2 Parameters for cases with failed 3% individual chamber deviation, 3-mm distance to agreement individual chamber, or 3% mean dose deviations during verification portal measurements
Carbon Proton
With range shifter
Without range shifter
Depth 0-50 mm
Depth 50-100 mm
Depth 100-150 mm
Depth 150-200 mm
Failed γ (3 mm/3%) criteria
4 4
42 6
14 8
15 2
13 0
4 0
5 1
mean dose instead of the usual median dose. Changing the prescription to median dose allowed the verification measurement to pass. Recently, planar arrays of ionization chambers have begun to be used for patient-specific QA in both x-ray and ion teletherapy. These arrays offer much larger numbers of chambers, allowing a more detailed observation of the dose distribution but only in 1 depth plane for a given measurement. The arrays are also simpler to set up compared with a large WP. These arrays are currently being commissioned for use at SPHIC and investigated with respect to the number of required measurement depths to ensure safe and accurate treatments.
entire 3-dimensional dose distribution delivered by teletherapy beams.16
Ethical Approval This article does not contain any studies with human participants or animals performed by any of the authors.
Informed Consent For this type of study, formal consent is not required.
Acknowledgments Conclusions The present study evaluated both standard cube plans and 1734 patient-specific portal verification measurements. Deviations between standard cube portal dose measurements and predicted planned doses were minimal. This finding was expected as these cube plans delivered uniform dose distributions. On the other hand, patient-specific verification plans delivered quite inhomogeneous dose distributions, resulting in slightly larger, but generally acceptable, deviations. Stratification of results demonstrated that, for most chambers, the deviations between the planned and measured doses were small during the present study period, with 2 chambers having a lower pass rate than the others. No significant differences between rooms or with time were observed for chambers in the same WP. There was a 1% decrease in mean measured dose for carbon ion portals when the chamber array position was at depths between 150 and 200 mm, which may indicate that, for carbon ion beams, scatter may not be estimated correctly in the TPS. A possible explanation for dose deviations for plans having small target volumes is that some chambers of the array may lie just outside the target or within large dose gradient regions (lateral or distal penumbra regions), and small chambers array setup errors can result in large deviations. This type of deviation, similar to that found in a University of Texas MD Anderson Cancer Center report,15 requires further investigation. In the future, ion chamber arrays with larger numbers of chambers should allow more sampling of the
The authors thank the following quality assurance team members who participated in the measurements of the cube and verification plans: Jingni Chen, Stefanie Kaess, Yongqiang Li, Lienchun Lin, Yinxiangzi Sheng, Sun Lining Sun, Weiwei Wang, and Jingfang Zhao. Appreciation is also expressed to 10 students studying for their master’s degree at Fudan University who assisted with the measurements as part of their clinical training.
References 1. Ando, K.; Kase, Y. Biological characteristics of carbon-ion therapy. Int. J. Radiat. Oncol. Biol. Phys. 85:715–28; 2009. 2. Suit, H.; DeLaney, T.; Goldberg, S.; et al. Proton vs. carbon ion beams in the definitive radiation treatment of cancer patients. Radiother. Oncol. 95:3–22; 2010. 3. Schardt, D.; Schulz-Ertner, D. Heavy-ion tumor therapy: Physical and radiobiological benefits. Med. Phys. 82:383–425; 2010. 4. Braeuer, M. Commissioning of the Siemens Beam Application and Monitoring System. PTCOG 47; 2008. Available at: https://www .ptcog.ch/archive/conference_p&t&v/PTCOG47/presentations/ 6_Meeting_Saturday/Session5/Braeuer_Ptcog47/Ptcog47_Sat _Braeuer_SiemensFinal.pdf. Accessed February 22, 2018. 5. Eickhoff, H.; Haberer, T. HICAT—The German hospital-based light ion cancer therapy project. Proceedings of EPAC 2004, Lucerne, Switzerland 290-294; 2004. 6. Møller, S.P.; Albrechsetsen, F.S. A novel proton and light ion synchrotron for particle therapy. Proceedings of EPAC 2006, Edinburgh, Scotland 2305-2307; 2006. 7. Møller, S.P.; Albrechsetsen, F.S. Accelerator systems for particle therapy proceedings of EPAC 2006. Edinburgh, Scotland 2302-2304; 2006.
ARTICLE IN PRESS 10
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
8. American Association of Physicists in Medicine. Physical aspects of quality assurance in radiation therapy. AAPM Report No 13. New York: AAPM; 1994. 9. Regan, M.; Stathakis, S. SU-E-T-350: Three year analysis using ionization chamber array for patient specific IMRT QA. Med. Phys. 39(6):3784; 2012. 10. Scarlat, F. Ionization chamber dosimetry for conventional and laser-driven clinical hadron beams. J. Biosci. Med. 3:8–17; 2015. 11. Heeg, P. Quality assurance at the heavy-ion therapy facility at GSI. Strahlenther. Onkol. 175(suppl. 2):36–8; 1999. 12. Haberer, T.; Debus, J. The Heidelberg Ion Therapy Center. Radiother. Oncol. 73:S186–90; 2004.
13. Karger, C.P.; Jäkel, O.; Hartmann, G.H. A system for three-dimensional dosimetric verification of treatment plans in intensity-modulated radiotherapy with heavy ions. Med. Phys. 26:2125–32; 1999. 14. Low, D.A.; Harms, W.B. A technique for the quantitative evaluation of dose distributions. Med. Phys. 25:656–61; 1998. 15. Zhu, X.R.; Poenisch, F. Patient-specific quality assurance for prostate cancer patients receiving spot scanning proton therapy using single-field uniform dose. Int. J. Radiat. Oncol. Biol. Phys. 81:552–9; 2012. 16. Varasteh, A.M. Quality assurance of carbon ion and proton beams: A feasibility study for using the 2D MatriXX detector. Phys. Med. 32:831–7; 2016.
Medical Dosimetry j o u r n a l h o m e p a g e : w w w. m e d d o s . o r g
Medical Physics Contribution:
Analysis of ion beam teletherapy patient-specific quality assurance Xiaoli Liu, M.S., Yu Deng, M.S., Nicki Schlegel, M.S., Zhijie Huang, Ph.D., and Michael F. Moyers, Ph.D. Department of Medical Physics, Shanghai Proton and Heavy Ion Center, Shanghai 201315, China
A R T I C L E
I N F O
Article history:
Received 25 April 2017 Accepted 25 January 2018 Keywords:
Quality assurance Ion beam Modulated scanning
A B S T R A C T
The objective of this study was to evaluate the procedures for patient-specific quality assurance measurements using modulated scanned and energy stacked beams for proton and carbon ion teletherapy. Delivery records from 1734 portal measurements were analyzed using a 3-point pass criteria: more than 22 of 24 chambers in a water phantom (WP) had to have a measured dose difference from the planned portal doses less than or equal to 3%, or the distance from the measurement point location to a point location in the plan having the same dose had to be less than or equal to 3 mm (distance to agreement [DTA]), and the mean dose deviation of all chambers had to be less than 3%. Stratification of results showed some associations between measurement parameters and pass rates. For proton portals, pass rates were high at all measurement depths, but for carbon ion portals, pass rates decreased as a function of increasing measurement depth. Pass rates of both proton and carbon ion portals with 1 WP were slightly lower than those with a second WP. The total pass rates were 97.7% and 91.9% for proton and carbon ion patient portals, respectively. In general, the measured doses exhibited good agreement with the treatment planning system (TPS) calculated doses. When the chamber position was deeper than 150 mm in carbon ion beams, a lower pass rate was observed, which may have been caused by ion chamber array setup uncertainty (lateral and depth) in highly modulated portals or incorrect modeling of scatter by the TPS. These deviations need further investigation. © 2018 American Association of Medical Dosimetrists.
Introduction Teletherapy beams of light ions, such as protons and carbons, share the physical advantage of a finite range in depth with a steep distal dose gradient after the so-called Bragg peak.1-3 By combining multiple energies to place the
Reprint requests to Xiaoli Liu, M.D., Department of Medical Physics, Shanghai Proton and Heavy Ion Center, Shanghai 201315, China. E-mail: [email protected]
Bragg peak at different depths throughout the target, a conformal dose distribution can be delivered to the target while minimizing radiation effects in the surrounding normal tissues. In the year 2014, the Shanghai Proton and Heavy Ion Center (SPHIC) opened and performed a successful clinical trial of patients using a Siemens Iontris treatment delivery system.4 This system provides modulated scanned and energy stacked proton and carbon ion beams with ranges in water up to 310 mm (energy per nucleons of 221 MeV/n for proton and 430 MeV/n for carbon ions)5-7; this system does not currently support the use of apertures or boluses. The SPHIC facility has 4 ion beam treatment rooms, 3 rooms with
https://doi.org/10.1016/j.meddos.2018.01.002 0958-3947/Copyright © 2018 American Association of Medical Dosimetrists
ARTICLE IN PRESS 2
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
horizontally oriented beamlines, and 1 room with a beamline at a 45° incline. All rooms are equipped with robotic patient positioners with an exchangeable flat table top mounted onto a robotic arm. Each treatment room also has an imaging robot mounted on the ceiling to verify patient alignment before treatment. Routine patient operations began after the clinical trial. Similar to the introduction of dynamic multileaf collimators into radiotherapy practice using megavoltage × ray beams, the move to modulated scanned and energy stacked ion beams also requires additional and efficient methods of patient-specific quality assurance (QA).8-10 Between 2015 and 2016, patient-specific QA was performed using a clustered array of ion chambers in a water phantom (WP). This report provides a summary and analysis of 1734 portal delivery records during the period from July 2015 to July 2016 during routine operations.
Fig. 1. Water phantom setup at isocenter for measurement of patientspecific portal. The inset at the upper right is a close-up picture of the ionization chamber array. (Color version of figure is available online.)
Methods and Materials At SPHIC, a QA procedure was performed for every patient portal before it was delivered to a patient. The current QA procedure was similar to that which was performed at the Gesellschaft für Schwerionenforschung and the Heidelberg Ion Therapy Center.11,12 This procedure consisted of copying all the portal parameters (spot positions, energies, numbers of ions, use of a beamline range shifter [RS], etc.) from each portal of a patient plan into a so-called verification plan that consisted of a rectangular parallelepiped WP, with each portal being directed into the side of the phantom. A verification plan may include more than 1 portal used for a given patient during a given treatment fraction, but in the verification plan, all portals entered the phantom from the same direction. The dose distribution of each portal was then recalculated, the verification plan was transmitted to the beam delivery system, and then all portals of this verification plan were delivered with an arbitrary horizontal beamline room. The use of any room for the verification measurement was possible because the beam range and beam monitors in all rooms were calibrated identically to within the clinical tolerance. The ion beam treatment planning system (TPS) used at SPHIC was Syngo RT Planning VC11B (Siemens, Erlangen, Germany). The WP input into the TPS for verification planning represented a real WP (PTW, Freiburg, Germany) that contained an array of 24 miniature thimble chambers in a polymethylmethacrylate block attached to a 3-dimensional scanning arm.13 The 24 chambers (PTW model 31015, PTW) each had an active volume of 0.03 cm3 and were arranged in 6 rows perpendicular to the beam axis (i.e., 6 depths), with each row containing 4 chambers. Every other row was offset laterally from the one in front, and every second row was offset slightly in the vertical direction to reduce perturbations
of the delivered dose distributions at chamber locations at deeper depths (see Fig. 1 inset). The chamber array covered a cubical volume of approximately 55 mm × 70 mm × 60 mm, and the planner could place this volume at selected locations within the WP to measure the absolute dose both in the center of the target and its periphery. In addition to providing the expected dose at each chamber location of the array, the TPS also provided the dose gradient at each chamber location to assist in the analysis of potential dose deviations. The dose calculations did not include the effects of the ion chamber air cavities or the effects of the polymethylmethacrylate block. Figure 1 shows the phantom setup in the treatment room. The facility utilized 2 separate but identical WPs and chamber arrays to measure patient portals, WP 1 and WP 2. Typically, only 1 phantom was used on a given day with the other one being used only for backup in case the primary one was not working properly or the chambers and the electrometer associated with the phantom had been sent for calibration. Because the WP entrance surface was always perpendicular to the floor, it could not be used with the oblique beamline (room 3); therefore, all measurements are performed in one of the horizontal rooms (rooms 1, 2, and 4). When routine clinical operations started, measurements were made at 3 different positions of the chamber array block for every portal, typically near the surface, in the middle of the target, and near the distal end of the target. After several months of operation, it was discovered that the pass rates for chamber locations near the surface were usually very high because the dose gradients in this region were typically quite small; thereafter, only 1 array block position was typically used and chosen to be located within the high-dose target volume. If the target was large, 2 positions were sometimes chosen, 1 in the middle of the target and 1 near the
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
3
Fig. 2. TPS display of a single portal of a verification plan. The upper left panel represents a view of the dose distribution in a vertical plane as seen from the side of the phantom. The upper right panel represents a view of the dose distribution in a horizontal plane as seen from the top of the phantom. The lower left panel represents a view of the dose distribution in a vertical plane perpendicular to the beam axis. The lower right panel is similar to the upper right panel but shows each ion chamber as a circle and a display of the absolute dose per fraction and dose gradient for each chamber. The color for each circle matches the isodose color levels. (Color version of figure is available online.)
distal end of the target to check the beam range. A TPS display of a single portal from a verification plan is shown in Fig. 2. The scanning WP was connected to the beam delivery system via an Ethernet connection to enable automatic setup of the array positions according to the verification plan. During measurement, the delivery system was run in a medical delivery mode with all beam delivery interlocks and tolerances identical to those used during treatment. To check the integrity of the WP acquisition system setup before each
measurement session of patient verification plans, standard plans for cubical targets were measured. Simultaneously, the dose per MU for the standard plans was checked. The standard plans for both proton and carbon ion beams were designed for cubical targets of 80 mm × 80 mm × 80 mm and a physical dose of 2 Gy per fraction. The depths for the center of the front row of chambers for the chamber array for the 5 standard proton cube plans were 36.3, 70.2, 125.2, 175.2, and 222.2 mm. The depths for the center of the front row of chambers for the chamber array for the 5 standard carbon
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
4
Table 1 Stratification of measurements
Carbon cubes Proton cubes Carbon portals Proton portals
Total
Water phantom 1
Water phantom 2
Depth 0-50 mm
Depth 50-100 mm
Depth 100-150 mm
Depth 150-200 mm
Room 1
Room 2
Room 4
136 109 1211 523
30 70 296 191
106 39 916 332
23 15 790 361
7 53 278 139
99 39 131 22
7 2 12 1
45 34 383 122
41 40 365 219
50 35 463 182
ion cube plans were 36.4, 71.3, 121.3, 171.3, and 221.3 mm. As seen in Table 1, cube plan 2 was most often used for proton measurement sessions and cube plan 3 was most often used for carbon ion measurement sessions. Patient-specific verification measurements were done in horizontal rooms using either WP 1 or WP 2. After portal delivery, the beam delivery system saved all the delivery and measurement records in an XML formatted file, including all beam parameters, the depth of the first row of chambers, and the beamline identification. The file was then imported into Excel for reading the information and performing analysis. From July 2015 to July 2016, a total of 1734 portal measurements (523 protons and 1211 carbon ions) from 608 plans were performed at SPHIC with some of these being repeat measurements.
Results and Discussion Table 1 shows the number of measurements for cube and patient verification plans stratified by WP, depth of
ion chamber block in phantom, and measurement room. Cubical target tests During the period of the present study, there were 245 cube measurements. Figure 3 shows the dose deviations of the cube measurements from their expected values stratified by the different categories described in the Methods and Materials section. In general, the measured cube doses matched the expected plan doses to better than ± 1%. The proton cube measurements showed smaller deviations than the carbon ion cube measurements. To investigate the behavior of each chamber of the array, the mean percentage dose deviations of each chamber for proton and carbon ion cube plans are plotted in Fig. 4. The mean percentage deviation for all chambers was within ± 2% with chambers 4 and 5 of WP 1 measuring the lowest doses for proton portals (− 1.56 ± 0.93% for chamber 4 and − 1.04 ± 0.74% for chamber 5). A similar phenomenon was found for proton measurements using WP 2.
Fig. 3. Mean percentage dose deviations for proton and carbon ion cube measurements stratified by WP, depth of ion chamber block in phantom, and measurement room. Plotted error bars represent 1 standard deviation from the mean. (Color version of figure is available online.)
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
5
Fig. 4. Mean percentage dose deviation for each chamber for proton and carbon ion cube measurements in water phantoms 1 and 2. Plotted error bars represent ± 1 standard deviation of each chamber. (Color version of figure is available online.)
Patient verification measurements The current criteria for a chamber measurement to pass are either that the deviation of the measured dose from the predicted planned dose is less than or equal to 3% or the distance between the location of a measured data point and the nearest point in the planned dose distribution that has the same dose is less than or equal to 3 mm (referred to as distance to agreement [DTA]). The portal verification is considered to pass if more than 22 of the 24 chambers pass these criteria and the mean dose deviation of all the chambers is less than or equal to 3%. During the study period, 523 proton portal measurements were performed. The absolute value of the mean percentage dose deviation between the measured and
planned doses for all chambers of the array was typically less than 1% for each measurement. For portals that included a RS, the mean of the mean dose deviation for all chambers of the array increased to − 1.19 ± 1.11%. Figure 5 shows the percentage mean dose deviations for patient verification measurements for proton portals stratified by WP, room, and chamber array depth. Figure 6 shows the percentage mean dose deviations for all proton portals as a function of ionization chamber. The mean deviation from the plan doses for most chambers was within ± 1%. The two chambers with the largest mean deviations were chamber 20 of WP 2 (+ 1.18 ± 3.05%) and chamber 5 of WP 1 (− 1.02 ± 2.13%). During the study period, 1211 carbon ion portal measurements were performed. Figure 7 shows the percentage
Fig. 5. Percentage mean dose deviation for proton portals stratified by WP, RS, depth of ion chamber array in phantom, and measurement room. Plotted error bars represent 1 standard deviation of the mean. (Color version of figure is available online.)
ARTICLE IN PRESS 6
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
Fig. 6. Percentage mean dose deviations of each chamber for all proton portals. Error bars represent 1 standard deviation of each chamber using water phantoms 1 and 2. (Color version of figure is available online.)
mean dose deviations for carbon ion portals stratified by WP, room, and chamber array depth. The absolute value of the percentage mean deviation between the measured and planned doses for all stratifications was less than 2.0%. For portals that included a RS, the percentage mean deviation was − 1.95 ± 1.46%. Figure 8 shows the percentage mean dose deviations for carbon ion portals as a function of ionization chamber. The absolute value of the percentage mean deviation for most chambers was within 2%. The 2 chambers with the largest mean deviations were chamber 5 of WP 1 (− 2.47 ± 2.30%) and chamber 20 of WP 2 (+ 0.09 ± 3.58%).
As seen in Figs. 6 and 8, the pass rates for proton and carbon ion ports were higher when using WP 2 than when using WP 1. This finding may be a result of the ion chamber calibration factors for phantom 2 being slightly higher than those for WP 1. The results also show that, on average, both proton and carbon ion portal measured doses were lower than the predicted planned doses. The total pass rates for proton and carbon ion portals were 97.7% and 91.9%, respectively. The percentage pass rates for all patient portals stratified by the parameters described previously are plotted in Fig. 9. The pass rate for carbon ion portals was seen to be a function of detector array depth,
Fig. 7. Percentage mean dose deviations for carbon ion portals stratified by WP, RS, depth of ion chamber array in phantom, and measurement room. Error bars represent 1 standard deviation of the mean. (Color version of figure is available online.)
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
7
Fig. 8. Percentage mean dose deviations of each chamber for carbon ion portals. Plotted error bars represent 1 standard deviation of each chamber using water phantoms 1 and 2. (Color version of figure is available online.)
with the lowest pass rate being observed when the chamber array depth was between 150 and 200 mm (66.7%). It should be noted, however, that this deepest stratification category contained only a few measurements (n = 13). This phenomenon needs further investigation. For proton portals, no significant correlation with depth was observed. The total pass rates for each of the individual chamber array positions in both WPs for all proton and carbon ion portal measurements are plotted in Fig. 10. The lowest pass rates for any chamber position were found in WP 1 for chambers 4 and 5 (89.2% and 83.4%, respectively) when used with the carbon ion beam. This phantom and chamber array was used infrequently but, in May 2016, a comprehensive check of the system revealed a slightly higher leakage rate for these 2 chamber channels. For portals with long delivery times
(>5 minutes), this finding may have resulted in a 1.6% to 2.0% lower reported dose, causing the chamber measurements to fail. There have been insufficient data taken after repairs by the manufacturer to evaluate if this leakage was indeed the cause of the lower pass rate. The total pass rates of portals from both patient-specific and standard cube plans as a function of time are plotted in Fig. 11. Before November 2015, WP 1 was the primary measurement device for verification plans. In November of 2015, the chambers for this phantom were sent for calibration and WP 2 was then used as the primary measurement device. Subsequent to this change, the pass rate for patientspecific portals was seen to increase. Out of a total of 1734 portal measurements from 608 plans (including some remeasured plans), 56 plans failed to pass
Fig. 9. Proton and carbon ion patient-specific portal pass rate stratified by WP, RS, and depth of ion chamber array in phantom. (Color version of figure is available online.)
ARTICLE IN PRESS 8
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
Fig. 10. Pass rates of each chamber position of the array for proton and carbon ion patient-specific portal measurements. (Color version of figure is available online.)
the criteria of 22 of 24 chambers having less than or equal to 3% dose deviations, 22 of 24 chambers having a DTA of less than or equal to 3 mm, and a mean dose deviation for all chambers less than 3%. A verification plan is considered to fail if any of the portals of that plan failed. Some plans that originally failed passed by remeasuring the failed portal or by shifting the chamber array to account for slight setup uncertainties. Forty-nine of the failed plans occurred for carbon ion plans and 7 for proton plans. Each of those failures required a more critical analysis. Table 2 summarizes the major plan parameters for those failures. Most of the failures were caused by large dose deviations of individual chambers. If the pass criteria were set to use a gamma index14 with a 3% dose difference or 3-mm DTA instead of the 3-point
criteria evaluation described above, then only 6 of the 608 plan measurements would have been considered to have failed. Four of those failed cases were for prostate treatments using carbon ion beams. These failed because of the mean dose deviation being larger than 3%. In these cases, the measurements were reperformed with the chamber array at slightly different locations, resulting in a pass for each case. One of the failed cases was for an adenoid cystic carcinoma treated with a carbon ion beam and a RS. This case also failed because of the mean dose deviation being larger than 3%. In this case, the plan was reoptimized, after which the verification measurement passed. Only 1 proton beam case was found to fail using the individual criteria test. Upon investigation, it was found that the prescription was set to the
Fig. 11. Percentage pass rate for both standard cube plans and patient-specific portal measurements as a function of time binned into monthly intervals. (Color version of figure is available online.)
ARTICLE IN PRESS X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
9
Table 2 Parameters for cases with failed 3% individual chamber deviation, 3-mm distance to agreement individual chamber, or 3% mean dose deviations during verification portal measurements
Carbon Proton
With range shifter
Without range shifter
Depth 0-50 mm
Depth 50-100 mm
Depth 100-150 mm
Depth 150-200 mm
Failed γ (3 mm/3%) criteria
4 4
42 6
14 8
15 2
13 0
4 0
5 1
mean dose instead of the usual median dose. Changing the prescription to median dose allowed the verification measurement to pass. Recently, planar arrays of ionization chambers have begun to be used for patient-specific QA in both x-ray and ion teletherapy. These arrays offer much larger numbers of chambers, allowing a more detailed observation of the dose distribution but only in 1 depth plane for a given measurement. The arrays are also simpler to set up compared with a large WP. These arrays are currently being commissioned for use at SPHIC and investigated with respect to the number of required measurement depths to ensure safe and accurate treatments.
entire 3-dimensional dose distribution delivered by teletherapy beams.16
Ethical Approval This article does not contain any studies with human participants or animals performed by any of the authors.
Informed Consent For this type of study, formal consent is not required.
Acknowledgments Conclusions The present study evaluated both standard cube plans and 1734 patient-specific portal verification measurements. Deviations between standard cube portal dose measurements and predicted planned doses were minimal. This finding was expected as these cube plans delivered uniform dose distributions. On the other hand, patient-specific verification plans delivered quite inhomogeneous dose distributions, resulting in slightly larger, but generally acceptable, deviations. Stratification of results demonstrated that, for most chambers, the deviations between the planned and measured doses were small during the present study period, with 2 chambers having a lower pass rate than the others. No significant differences between rooms or with time were observed for chambers in the same WP. There was a 1% decrease in mean measured dose for carbon ion portals when the chamber array position was at depths between 150 and 200 mm, which may indicate that, for carbon ion beams, scatter may not be estimated correctly in the TPS. A possible explanation for dose deviations for plans having small target volumes is that some chambers of the array may lie just outside the target or within large dose gradient regions (lateral or distal penumbra regions), and small chambers array setup errors can result in large deviations. This type of deviation, similar to that found in a University of Texas MD Anderson Cancer Center report,15 requires further investigation. In the future, ion chamber arrays with larger numbers of chambers should allow more sampling of the
The authors thank the following quality assurance team members who participated in the measurements of the cube and verification plans: Jingni Chen, Stefanie Kaess, Yongqiang Li, Lienchun Lin, Yinxiangzi Sheng, Sun Lining Sun, Weiwei Wang, and Jingfang Zhao. Appreciation is also expressed to 10 students studying for their master’s degree at Fudan University who assisted with the measurements as part of their clinical training.
References 1. Ando, K.; Kase, Y. Biological characteristics of carbon-ion therapy. Int. J. Radiat. Oncol. Biol. Phys. 85:715–28; 2009. 2. Suit, H.; DeLaney, T.; Goldberg, S.; et al. Proton vs. carbon ion beams in the definitive radiation treatment of cancer patients. Radiother. Oncol. 95:3–22; 2010. 3. Schardt, D.; Schulz-Ertner, D. Heavy-ion tumor therapy: Physical and radiobiological benefits. Med. Phys. 82:383–425; 2010. 4. Braeuer, M. Commissioning of the Siemens Beam Application and Monitoring System. PTCOG 47; 2008. Available at: https://www .ptcog.ch/archive/conference_p&t&v/PTCOG47/presentations/ 6_Meeting_Saturday/Session5/Braeuer_Ptcog47/Ptcog47_Sat _Braeuer_SiemensFinal.pdf. Accessed February 22, 2018. 5. Eickhoff, H.; Haberer, T. HICAT—The German hospital-based light ion cancer therapy project. Proceedings of EPAC 2004, Lucerne, Switzerland 290-294; 2004. 6. Møller, S.P.; Albrechsetsen, F.S. A novel proton and light ion synchrotron for particle therapy. Proceedings of EPAC 2006, Edinburgh, Scotland 2305-2307; 2006. 7. Møller, S.P.; Albrechsetsen, F.S. Accelerator systems for particle therapy proceedings of EPAC 2006. Edinburgh, Scotland 2302-2304; 2006.
ARTICLE IN PRESS 10
X. Liu et al. / Medical Dosimetry ■■ (2018) ■■–■■
8. American Association of Physicists in Medicine. Physical aspects of quality assurance in radiation therapy. AAPM Report No 13. New York: AAPM; 1994. 9. Regan, M.; Stathakis, S. SU-E-T-350: Three year analysis using ionization chamber array for patient specific IMRT QA. Med. Phys. 39(6):3784; 2012. 10. Scarlat, F. Ionization chamber dosimetry for conventional and laser-driven clinical hadron beams. J. Biosci. Med. 3:8–17; 2015. 11. Heeg, P. Quality assurance at the heavy-ion therapy facility at GSI. Strahlenther. Onkol. 175(suppl. 2):36–8; 1999. 12. Haberer, T.; Debus, J. The Heidelberg Ion Therapy Center. Radiother. Oncol. 73:S186–90; 2004.
13. Karger, C.P.; Jäkel, O.; Hartmann, G.H. A system for three-dimensional dosimetric verification of treatment plans in intensity-modulated radiotherapy with heavy ions. Med. Phys. 26:2125–32; 1999. 14. Low, D.A.; Harms, W.B. A technique for the quantitative evaluation of dose distributions. Med. Phys. 25:656–61; 1998. 15. Zhu, X.R.; Poenisch, F. Patient-specific quality assurance for prostate cancer patients receiving spot scanning proton therapy using single-field uniform dose. Int. J. Radiat. Oncol. Biol. Phys. 81:552–9; 2012. 16. Varasteh, A.M. Quality assurance of carbon ion and proton beams: A feasibility study for using the 2D MatriXX detector. Phys. Med. 32:831–7; 2016.