- Email: [email protected]
Feasibility Study of Glass Dosimeter for In Vivo Measurement: Dosimetric Characterization and Clinical Application in Proton Beams
International Journal of
Radiation Oncology biology
physics
www.redjournal.org
Physics Contribution
Feasibility Study of Glass Dosimeter for In Vivo Measurement: Dosimetric Characterization and Clinical Application in Proton Beams Jeong-Eun Rah, PhD,* Do Hoon Oh, MD,* Jong Won Kim, MS,* Dae-Hyun Kim, MS,y Tae-Suk Suh, PhD,y Young Hoon Ji, PhD,z Dongho Shin, PhD,x Se Byeong Lee, PhD,x Dae Yong Kim, MD,x and Sung Yong Park, PhDjj *Department of Radiation Oncology, Myongji Hospital, Kwandong University College of Medicine, Goyang, Korea; y Department of Biomedical Engineering, The Catholic University of Korea, Seoul, Korea; zKorea Institute of Radiological and Medical Sciences, Seoul, Korea; xProton Therapy Center, National Cancer Center, Goyang, Korea; and jjProton Therapy Center, McLaren Cancer Institute, Flint, Michigan Received May 17, 2011, and in revised form Feb 14, 2012. Accepted for publication Mar 23, 2012
Summary We evaluated the dosimetric characteristics of the glass dosimeter for in vivo dosimetry in a proton beam and concluded that the glass dosimeter has considerable potential as a new dosimeter using a clinical proton beam.
Purpose: To evaluate the suitability of the GD-301 glass dosimeter for in vivo dose verification in proton therapy. Methods and Materials: The glass dosimeter was analyzed for its dosimetrics characteristic in proton beam. Dosimeters were calibrated in a water phantom using a stairlike holder specially designed for this study. To determine the accuracy of the glass dosimeter in proton dose measurements, we compared the glass dosimeter and thermoluminescent dosimeter (TLD) dose measurements using a cylindrical phantom. We investigated the feasibility of the glass dosimeter for the measurement of dose distributions near the superficial region for proton therapy plans with a varying separation between the target volume and the surface of 6 patients. Results and Discussion: Uniformity was within 1.5%. The dose-response has good linearity. Dose-rate, fading, and energy dependence were found to be within 3%. The beam profile measured using the glass dosimeter was in good agreement with the profile obtained from the ionization chamber. Depth-dose distributions in nonmodulated and modulated proton beams obtained with the glass dosimeter were estimated to be within 3%, which was lower than those with the ionization chamber. In the phantom study, the difference of isocenter dose between the delivery dose calculated by the treatment planning system and that measured by the glass dosimeter was within 5%. With in vivo dosimetry, the calculated surface doses overestimated measurements by 4%-16% using glass dosimeter and TLD. Conclusion: It is recommended that bolus be added for these clinical cases. We also believe that the glass dosimeter has considerable potential for use with in vivo patient proton dosimetry. Ó 2012 Elsevier Inc.
Reprint requests to: Sung Yong Park, PhD, McLaren Cancer Institute, McLaren-Flint, 4100 Beecher Rd, Suite A, Flint, MI 48532. Tel: (810) 342-3849; Fax: (810) 342-3784; E-mail: [email protected] or [email protected] Int J Radiation Oncol Biol Phys, Vol. 84, No. 2, pp. e251ee256, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2012.03.054
Conflict of interest: none. AcknowledgmentdThe authors would like to thank Nick Stanley and Amy Nolin for proof reading this manuscript and this work was supported by a grant from the National Cancer Center (No.1110600) in Korea.
e252 Rah et al.
International Journal of Radiation Oncology Biology Physics
Introduction
maximum height was 100 mm, with each step of 1 mm in 10 layers. Delivery dose to the dosimeters was 2 Gy at the center of the spread-out Bragg peak (SOBP) in a water phantom with a modulated 200 MeV proton beam. Water level was adjusted precisely at a surface depth from the effective readout center of the dosimeters. Absorbed dose to water in the proton beam was determined using a Farmer type ionization chamber according to the IAEA Code of Practice TRS-398 (10).
In vivo dosimetry is important for verification of the actual dose delivered to the target volume and for estimation of dose to organs at risk. A number of studies have attempted to evaluate the accuracy of such dose measurements in proton beam for in vivo dosimetry with new dosimetric materials as well as with conventional dosimeters (1-6). In addition, as more proton therapy facilities are being established, the use of high-energy proton beams in radiation therapy is becoming a topic of increasing interest. Although dosimetric characteristics with various types of dosimeters for high-energy proton beams have been reported in the literature, most of their results demonstrated the basic limitation of dosimetry for a pristine Bragg peak, particularly in the distal falloff region. The main problem in these cases is the decrease in the response of the dosimeter with increasing ionization density of the radiation field. This may lead to underestimation of dose after heavy charged particle irradiation. In recent years, use of a commercially available radiophotoluminescent glass dosimeter system for radiation measurement has increased (7-9). In a previous study, we evaluated the dosimetric characteristics of the glass dosimeter to examine whether they would be suitable for in vivo dose verification in high-energy photon beam (9). Results showed the glass dosimeter can be used as an accurate and reproducible dosimeter for skin dose measurements. The purpose of this study was to evaluate the suitability of the glass dosimeter for in vivo dose verification in proton therapy. First, we studied dosimetric characteristics of the glass dosimeter with regard to response uniformity, dose linearity, dose-rate dependence, fading effect, beam profile, depth-dose distribution, and energy dependence. Second, we compared the dose measured with the glass dosimeter and calculated the dose of treatment planning system (TPS) using a cylindrical phantom. Finally, we investigated the feasibility of the glass dosimeter for measurement of dose distributions near the superficial region of 6 patients for proton therapy plans.
Methods and Materials Glass dosimeter system For our study, a model GD-301 glass dosimeter (AGC Techno Glass, Shizuoka, Japan) and FGD-1000 automatic reader were used. Model GD-301 is 1.5 mm in diameter and 8.5 mm in length; the effective readout size is 1 mm in diameter and 0.6 mm in depth (7). Effective atomic number and density of the glass dosimeter were 12.039 and 2.61 g/cm3, respectively. The glass dosimeter was annealed (at 400 C for 1 hour) before it was used for measurement.
Dosimetric characteristics of the proton beam The proton beam reference condition in this study consists of a snout size with an 18 cm diameter and a 10 10 cm2 open block. Uniformity in the response of the glass dosimeters was evaluated for 50 dosimeters with the 200-MeV proton beam in a water phantom. Delivery dose to the dosimeter was approximately 2 Gy at the middle of the SOBP. Dose-linearity evaluates response as a function of the dose delivered to each dosimeter. For determination of their linearity, dosimeters were irradiated with a proton beam in the dose range of 1-10 Gy for the reference condition. For dose-rate dependence, glass dosimeters were irradiated to approximately 2 Gy at each of the following dose rates: 50, 100, 200, 300, 400, and 500 MU/min. For determination of the change in the measured dose with delayed readout time, fading characteristics were investigated under normal ambient conditions during a storage period ranging from 5-60 min postirradiation. Beam profile, which can be obtained from glass dosimeters, was oriented with its axis perpendicular to the beam axis. We irradiated the glass dosimeters with a modulated 160 MeV proton beam. The dosimeters were positioned in a polystyrene plate phantom with a hole. A dose of 2 Gy was delivered to the glass dosimeters at the center of the SOBP. The result measured with the glass dosimeter was evaluated by comparison with those from a plane-parallel Markus type ionization chamber (PTW, TB23343). The diameter and height of the measuring volume are 5 mm and 1 mm, respectively. Depth-dose measurements were 8.0 cm proton range in water, where the mean energy was approximately 160 MeV for both nonmodulated and modulated beams. Measurements were performed in a water phantom using a stairlike holder. The dosimeter was aligned with the center of the beam axis at the isocenter using laser lights. For comparison, depth dose distribution was also measured by the Markus chamber. To determine the varying responses of the glass dosimeter, with proton beam energy, the glass dosimeter was irradiated to a dose of 2 Gy for proton ranges between 5.0 and 19.1 cm. For a given range, a glass dosimeter was placed in a water phantom using a stairlike holder corresponding to the center of the SOBP.
Phantom study Calibration of the glass dosimeter All measurements were performed in a clinical proton beam (IBA Proton Therapy System-Proteus 235) at the National Cancer Center in Korea. The minimum and maximum ranges of the proton beam in patients are 5 cm and 28 cm, respectively, with 0.1 cm accuracy in the double-scattering mode. Glass dosimeters were calibrated by delivery of a known dose in a water phantom using a stairlike holder specially designed for this study. The
The proton dose verification was delivered to an in-house manufactured phantom that was constructed with polymethylmethacrylate, was cylindrical in shape, had an interchangeable plate insert for the dosimeter, was 25 cm in diameter and 20 cm in length. We compared the glass dosimeter and thermoluminescent dosimeter (TLD) dose measurements. The TLDs used in this study were Harshaw TLD-100 lithium fluoride chips with a nominal thickness of 1 mm and a normal surface area of 3 3 mm2.
Volume 84 Number 2 2012
Fig. 1.
In vivo dosimetry with glass dosimeter in proton beams e253
Dose verification measurement using the cylindrical phantom for a proton therapy plan.
The dosimeter evaluation was assessed using 3 separate sets of measurement methods. Dose calculations in Eclipse (Varian Medical Systems, Palo Alto, CA) were performed with a grid size of 2.5 mm. For the first measurement, the superficial planning target volume (PTV) was created by autocontouring of the entire surface of the phantom (Fig. 1a). The surface doses were measured by placing the dosimeters on the front surface of the phantom. The delivered dose for the PTV was 2 Gy with 1 beam angle. In a second set of measurements, the planning target volume (PTV) was localized to the center of the phantom. Dosimeters were then placed at central and surface
points on the phantom, as seen in Fig. 1b. The plan was optimized to give a homogeneous dose of 2 Gy per fraction to the target region with the anterior-posterior beam. Finally, dosimeter measurements were performed in 2 clinical cases with different treatment sites. Figure 1c and 1d shows dose distributions for a proton plan generated by Eclipse. One case was planned for brain cancer with the PTV prescribed to 54 Gy at 2 Gy per fraction. The other case was treated for liver cancer with the PTV prescribed to 55 Gy at 2.5 Gy per fraction. Each experiment was repeated 5 times, and an average measurement was used.
Fig. 2. Dose distribution of treatment proton plans. (a and b) Transverse and sagittal view of patient 1 (breast cancer) without the use of bolus. (c and d) Transverse and sagittal view of patient 3 (inguinal cancer) with the use of bolus.
e254 Rah et al.
International Journal of Radiation Oncology Biology Physics
Because the glass dosimeter and TLD used in this study had a physical thickness of 1.5 mm and 1 mm, respectively, these measurements overestimate the entrance dose. It is important to note that the dosimeters dose results reported in this study are not true surface dose measurements. Therefore, to account for any uncertainty in the positioning of the dosimeters, the positioning of the phantom and the error from the 2.5 mm dose grid, the dosimeter region of interest was extended by 1 voxel outside the visible dosimeters locations. The mean dose and the standard deviation were obtained from calculated dose values bound within the contoured volume.
Patient study Glass dosimeters were used on 6 patients for in vivo dose measurements. We selected patients with clinical localization near the superficial target, which were treated with proton radiation therapy at our institution. The proton therapy cases were planned using the Eclipse for all patients. Figure 2 shows the treatment plan for breast cancer (patient 1) and anal cancer (patient 2), respectively, along with the dose distribution for the corresponding coronal and axial views. Patients 3 and 4 were treated by adding a 10-mm bolus to increase the skin dose. One fraction was delivered and measured with a glass dosimeter and TLD-100 chip for each patient. The dosimeters were arranged into 2 packets, with 3 glass dosimeters and TLD chips per packet on the patient’s skin.
Results and Discussion Dosimetric characteristics of the proton beam Response uniformity was estimated by taking 1 standard deviation of the average for the glass dosimeter. The uniformity result was within 1.5%. Unlike the TLD, which can only be read once, the radiophotoluminescence (RPL) signal does not disappear after the reading operation. Therefore, the glass dosimeter can be read an unlimited number of times (7-10). This provides an advantage in performance of in vivo dosimetry on patients, which requires accurate measurement. The glass dosimeter was found to have a good linear relationship for a dose ranging from 1-10 Gy with a linearity coefficient R of 0.999. The dose-rate dependence was negligible for in vivo purposes and was within 1.5%. The fading effect of the glass dosimeter was relatively stable, within 1.5% for a time delay of 60 min. In our previously published study, the glass dosimeter also presented long-term stability due to less fading. After 150 days of storage, losses in the response of the glass dosimeter did not exceed 2% (8). Figure 3a shows the comparison of beam profiles as measured by the glass dosimeter and Markus chamber in the horizontal plan at the center of the SOBP. It shows the flatness of the proton beam, as measured by the glass dosimeter, to be 1.06 in the horizontal plane. For the ionization chamber, the flatness was 1.02 in the horizontal plane. The beam penumbra between 80% and 20%, obtained from glass dosimeter, was 7.3 mm and was 7.0 mm with the ionization chamber. The profiles from both dosimeters are in agreement at an open block of 10 10 cm2. A comparison of depth-dose distribution for a 160-MeV nonmodulated and modulated proton beam obtained with the glass
Fig. 3. (a) Comparison of beam profile as measured with the glass dosimeter and Markus ionization chamber in horizontal plane. The profiles are normalized at the central axis of a beam. (b and c) Comparison of the Bragg curves obtained with the glass dosimeter and Markus ionization chamber for unmodulated and modulated proton beam. All data are normalized at beam entrance. dosimeters and the ionization chamber is presented in Fig. 3b and 3c. The ratio of the dose measured in the peak to the dose in the entrance region of the Bragg peak was 3.00 with the ionization chamber and 2.93 with the glass dosimeter; it was found to be within 2.5%. This shift was probably due to setup uncertainty in the glass dosimeter positioning of the effective measurement point in a water phantom using a stairlike holder. The range measured with the glass dosimeter was estimated to be 1.9% smaller than
Volume 84 Number 2 2012 Table 1
Set 1 Set 2 Set 3
In vivo dosimetry with glass dosimeter in proton beams e255
Results of the phantom study Measurement position
No. of irradiation fields
Calculation dose (Gy)
Surface Surface Isocenter Case 1: Brain isocenter Case 2 : Liver isocenter
1 1
2.00 1.40 2.00 2.05 2.48
4 3
Glass dosimeter
TLD
Mean (Gy) % SD
Mean (Gy) % SD
1.91 1.29 2.08 1.98 2.41
1.7 1.4 1.2 1.6 1.3
1.84 1.31 1.96 1.91 2.57
2.2 2.9 2.7 2.1 2.3
Abbreviations: SD Z standard deviation; TLD Z thermoluminescent dosimeter.
those from the ionization chamber. The depth-dose distribution of the spread-out Bragg curve obtained with the glass dosimeter was approximately 3% lower than with the ionization chamber at the end of the plateau. However, our results show that the glass dosimeter is more stable and relatively independent of linear energy transfer (LET) than the MOSFET and scintillator dosimeter, especially for large values of dE/dX at lower energies. Kohno et al (3) used the MOSFET for evaluation of characteristics for therapeutic proton beams. The Bragg peak obtained by the MOSFET was estimated to be approximately 40% lower than those by the ionization chamber, demonstrating that MOSFET responses are strongly dependent on LET for proton beams. Torrisi et al (5) evaluated the use of a plastic scintillator dosimeter in a proton beam; however, a plastic scintillator has never been used extensively for proton dose measurements. The main disadvantage of using scintillation detectors in proton therapy is the well-known quenching of the scintillator. Energy dependence of the glass dosimeter is within 3% for proton beam ranging from 5 cm-19.1 cm. Zullo et al (1) reported that energy dependence of Lithium Fluoride TLD powder for proton beams from 100-250 MeV agree within 5% of the actual dose values when the TLD was irradiated in a region of relatively uniform LET values. A recent article by Reft (4) reported on energy dependence of a carbon-doped aluminum oxide (Al2O3:C) as an OSL detector for proton beams. These results demonstrate that OSL response was found to be independent of the proton energy from 150-250 MeV and approximately 6% greater than their response to a 6-MV photon beam. The glass dosimeter is therefore well suited to surface dose measurements. They could be used to measure the entrance dose at the skin precisely.
Phantom study Results of 3 sets of glass dosimeter measurements for plan verification using the cylindrical phantom are summarized in Table 1. Table 2
Each point represents the average value of 5 measurements. Calibration of dosimeters was carried out at the center of the SOBP in a water phantom with a modulated 200 MeV proton beam. In the first and second set of measurements, the difference between the glass dosimeter measured dose and the calculated treatment plan dose at each surface point was 4.7% and 8.5%, respectively. This large deviation was investigated by review of the surface dose calculations performed by the Eclipse TPS (11). Surface dose points were estimated by reading the calculated doses at the intersection of the external contour and air. This boundary was chosen as the depth where the Hounsfield number changes from negative to positive. Since the dose calculation voxels were 2.5 mm3, some averaging takes place. Consequently, surface dose points extracted from the TPS also include uncertainties. The dose assigned to voxels near the air-surface interface was found to be significantly affected by the partial volume effect, leading to inaccurate dose estimation at points on the top of the phantom. On the other hand, for the second and third set, differences of isocenter dose between the calculation from Eclipse and measurement using the glass dosimeter are within approximately 5%. Concurrent TLD measurement differs from calculation by approximately 8%.
Patient study Calculated and measured surface doses delivered to patients are shown in Table 2. The difference between the glass dosimeter measured dose and the calculated treatment plan dose was as large as 11%, with an average of 5.9% 1.8%. Overall average agreement with the TLD and glass dosimeter measurement was within 2%. For surface dose, agreement with the treatment plan doses was similar to the results found with the phantom study. Doses measured by the TLD indicated that the proton therapy Eclipse TPS overestimates surface doses by 3%-16%, except for in patient 4, for whom it underestimated these doses by approximately 2.4%. Doses calculated and measured using the glass
Patient characteristics, prescription dose, and results of the patient study
Glass dosimeter TLD No. of treatment Prescribed Calculation Pt. no. Treatment site Age (y)/sex fields Bolus dose to PTV dose (Gy) Mean (Gy) % SD Mean (Gy) % SD 1 2 3 4 5 6
Breast Anal Inguinal Brain Rib Mediastium
53/F 27/F 54/F 37/M 64/M 71/M
2 2 2 3 3 3
þ þ -
2.00 2.00 2.00 2.20 3.00 2.40
1.89 1.96 1.85 2.07 2.92 2.17
1.79 1.77 1.78 2.00 2.82 2.01
1.9 1.8 1.6 1.9 1.8 1.4
1.82 1.72 1.71 2.12 2.75 1.88
2.8 2.6 2.1 2.7 3.2 2.5
Abbreviations: F Z female; M Z male; PTV Z planning target volume; SD Z standard deviation; TLD Z thermoluminescent dosimeter.
International Journal of Radiation Oncology Biology Physics
e256 Rah et al. Table 3 Uncertainty analysis for measured results obtained with glass dosimeters Physical quantity of procedure Step 1: Calibration procedure Dose measured with the ionization chamber Reading uniformity Setup error Combined uncertainty Step 2: Dose correction factor Combined uncertainty (dose rate, fading, energy dependence, etc) Step 3: In vivo dosimetry Combined uncertainty Overall uncertainty
Uncertainty (%)
measurement. In addition to the uniformity of in vivo dosimeters, components of this step establish the patient setup of conditions for clinical treatment. Therefore, we determined that the uncertainty in surface dose measurement with the glass dosimeter in the proton beam is approximately 4.3%.
2.0 1.2 0.5 2.4 2.9
2.0 4.3%
All uncertainties were estimated as 1 standard deviation.
dosimeter for clinical cases in which the target volume extends to the surface of the patient indicate that the target was underdosed by an average of 7% without the use of bolus. On the other hand, for patients 3 and 4, with use of a bolus to cover the treatment site, calculated values with Eclipse overestimated the measured dose by approximately 3.7%. As such, addition of bolus for these clinical cases is recommended. For patient 2, without the use of bolus, given the result of the measurements using the glass dosimeter and TLD-100, calculated doses on the surface of the patient were overestimated by between 10.7% and 14%. In accordance with our results, the treatment plan for patient 2 was modified for the second fraction to increase the tumor volume coverage of the prescribed dose. The PTV was recontoured including bolus material as a target region. The tumor volume coverage can be achieved with a surface dose to 95% of the prescribed dose. These results confirmed the inaccuracy of the calculation of surface doses by the TPS. Inaccurate calculations in the air-surface interface region will also affect the doses given to the normal skin tissues and near-superficial tumors. Therefore, it is important to measure the dose for near-surface tumors and also to measure the dose delivered to the skin.
Uncertainty analysis Table 3 shows the uncertainty analysis of the glass dosimeter. Evaluation of these uncertainties is important in experimental investigations. The combined standard uncertainty at each step is assessed by the square root of the quadratic sum of the individual uncertainties in dose measurement using the glass dosimeter. For the first step (ie, dose measurement with the reference ionization chamber), typical uncertainties in the clinical proton beam was approximately 2%. For calibration, setup uncertainty was based on the positioning accuracy in the phantom and the accuracy of the source-to-dosimeter distance. This component is also included in daily variations, machine setup, and machine instability during several months of calibration. Uncertainty of the second step could be regarded as the accuracy with which we could determine the correction factors for dosimetric characteristics of the proton beam. The uncertainty of the third step deals with in vivo
Conclusion We evaluated the dosimetric characteristics of the glass dosimeter for in vivo dosimetry in proton beams. Depth-dose distributions in nonmodulated and modulated proton beams obtained with the glass dosimeter were within 3%, lower than those with the ionization chamber. In the phantom study, the difference of isocenter dose between delivery dose calculated by the Eclipse and that measured by the glass dosimeter was within 5% when the dosimeter was calibrated at the center of the SOBP with the modulated proton beam. In vivo dosimetry, the calculated surface doses overestimated measurements by 4%-16% using the glass dosimeter and TLD. It is recommended that a bolus be used on the surface to ensure that tumors of superficial regions are not underdosed. From these results, we conclude that the glass dosimeter has considerable potential as a new dosimeter using a clinical proton beam.
References 1. Zullo JR, Kudchadker RJ, Zhu XR, et al. LiF TLD-100 as a dosimeter in high energy proton beam therapydcan it yield accurate results? Med Dosim 2009;35:63-66. 2. Grusell E, Medin J. General characteristics of the use of silicon diode detectors for clinical dosimetry in proton beams. Phys Med Biol 2000; 45:2573-2582. 3. Kohno R, Nishio T, Miyagishi T, et al. Experimental evaluation of a MOSFET dosimeter for proton dose measurements. Phys Med Biol 2006;51:6077-6086. 4. Reft CS. The energy dependence and dose response of a commercial optically stimulated luminescent detector for kilovoltage photon, megavoltage photon, and electron, proton, and carbon beams. Med Phys 2009;36:1690-1699. 5. Torrisi L. Plastic scintillator investigations for relative dosimetry in proton-therapy. Nucl Instrum Methods Phys Res 2000;170:523-530. 6. Daftari I, Castenadas C, Petti PL, et al. An application of GafChromic MD-55 film for 67.5 MeV clinical proton beam dosimetry. Phys Med Biol 1999;44:2735-2745. 7. Araki F, Ikegami T, Ishidoya T, et al. Measurements of gamma-knife helmet output factors using a radiophotoluminescent glass rod dosimeter and a diode detector. Med Phys 2003;30:1976-1981. 8. Rah J, Hong J, Kim G, et al. A comparison of the dosimetric characteristics of a glass rod dosimeter and a thermoluminescent dosimeter for mailed dosimeter. Radiat Meas 2009;44:18-22. 9. Rah J, Hwang U, Jeong H, et al. Clinical application of glass dosimeter for in vivo dose measurements of total body irradiation treatment technique. Radiat Meas 2011;46:40-45. 10. Absorbed Determination in External Beam Radiotherapy: an International Code of Practice for Dosimetry Based on Standard Absorbed Dose to Water [Technical Report Series No. 398]. Vienna: International Atomic Energy Agency; 2000. 11. Tournel K, Verellen D, Duchateau, et al. An assessment of the use of skin flashes in helical tomotherapy using phantom in-vivo dosimetry. Radiother Oncol 2007;84:34-39.
Radiation Oncology biology
physics
www.redjournal.org
Physics Contribution
Feasibility Study of Glass Dosimeter for In Vivo Measurement: Dosimetric Characterization and Clinical Application in Proton Beams Jeong-Eun Rah, PhD,* Do Hoon Oh, MD,* Jong Won Kim, MS,* Dae-Hyun Kim, MS,y Tae-Suk Suh, PhD,y Young Hoon Ji, PhD,z Dongho Shin, PhD,x Se Byeong Lee, PhD,x Dae Yong Kim, MD,x and Sung Yong Park, PhDjj *Department of Radiation Oncology, Myongji Hospital, Kwandong University College of Medicine, Goyang, Korea; y Department of Biomedical Engineering, The Catholic University of Korea, Seoul, Korea; zKorea Institute of Radiological and Medical Sciences, Seoul, Korea; xProton Therapy Center, National Cancer Center, Goyang, Korea; and jjProton Therapy Center, McLaren Cancer Institute, Flint, Michigan Received May 17, 2011, and in revised form Feb 14, 2012. Accepted for publication Mar 23, 2012
Summary We evaluated the dosimetric characteristics of the glass dosimeter for in vivo dosimetry in a proton beam and concluded that the glass dosimeter has considerable potential as a new dosimeter using a clinical proton beam.
Purpose: To evaluate the suitability of the GD-301 glass dosimeter for in vivo dose verification in proton therapy. Methods and Materials: The glass dosimeter was analyzed for its dosimetrics characteristic in proton beam. Dosimeters were calibrated in a water phantom using a stairlike holder specially designed for this study. To determine the accuracy of the glass dosimeter in proton dose measurements, we compared the glass dosimeter and thermoluminescent dosimeter (TLD) dose measurements using a cylindrical phantom. We investigated the feasibility of the glass dosimeter for the measurement of dose distributions near the superficial region for proton therapy plans with a varying separation between the target volume and the surface of 6 patients. Results and Discussion: Uniformity was within 1.5%. The dose-response has good linearity. Dose-rate, fading, and energy dependence were found to be within 3%. The beam profile measured using the glass dosimeter was in good agreement with the profile obtained from the ionization chamber. Depth-dose distributions in nonmodulated and modulated proton beams obtained with the glass dosimeter were estimated to be within 3%, which was lower than those with the ionization chamber. In the phantom study, the difference of isocenter dose between the delivery dose calculated by the treatment planning system and that measured by the glass dosimeter was within 5%. With in vivo dosimetry, the calculated surface doses overestimated measurements by 4%-16% using glass dosimeter and TLD. Conclusion: It is recommended that bolus be added for these clinical cases. We also believe that the glass dosimeter has considerable potential for use with in vivo patient proton dosimetry. Ó 2012 Elsevier Inc.
Reprint requests to: Sung Yong Park, PhD, McLaren Cancer Institute, McLaren-Flint, 4100 Beecher Rd, Suite A, Flint, MI 48532. Tel: (810) 342-3849; Fax: (810) 342-3784; E-mail: [email protected] or [email protected] Int J Radiation Oncol Biol Phys, Vol. 84, No. 2, pp. e251ee256, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2012.03.054
Conflict of interest: none. AcknowledgmentdThe authors would like to thank Nick Stanley and Amy Nolin for proof reading this manuscript and this work was supported by a grant from the National Cancer Center (No.1110600) in Korea.
e252 Rah et al.
International Journal of Radiation Oncology Biology Physics
Introduction
maximum height was 100 mm, with each step of 1 mm in 10 layers. Delivery dose to the dosimeters was 2 Gy at the center of the spread-out Bragg peak (SOBP) in a water phantom with a modulated 200 MeV proton beam. Water level was adjusted precisely at a surface depth from the effective readout center of the dosimeters. Absorbed dose to water in the proton beam was determined using a Farmer type ionization chamber according to the IAEA Code of Practice TRS-398 (10).
In vivo dosimetry is important for verification of the actual dose delivered to the target volume and for estimation of dose to organs at risk. A number of studies have attempted to evaluate the accuracy of such dose measurements in proton beam for in vivo dosimetry with new dosimetric materials as well as with conventional dosimeters (1-6). In addition, as more proton therapy facilities are being established, the use of high-energy proton beams in radiation therapy is becoming a topic of increasing interest. Although dosimetric characteristics with various types of dosimeters for high-energy proton beams have been reported in the literature, most of their results demonstrated the basic limitation of dosimetry for a pristine Bragg peak, particularly in the distal falloff region. The main problem in these cases is the decrease in the response of the dosimeter with increasing ionization density of the radiation field. This may lead to underestimation of dose after heavy charged particle irradiation. In recent years, use of a commercially available radiophotoluminescent glass dosimeter system for radiation measurement has increased (7-9). In a previous study, we evaluated the dosimetric characteristics of the glass dosimeter to examine whether they would be suitable for in vivo dose verification in high-energy photon beam (9). Results showed the glass dosimeter can be used as an accurate and reproducible dosimeter for skin dose measurements. The purpose of this study was to evaluate the suitability of the glass dosimeter for in vivo dose verification in proton therapy. First, we studied dosimetric characteristics of the glass dosimeter with regard to response uniformity, dose linearity, dose-rate dependence, fading effect, beam profile, depth-dose distribution, and energy dependence. Second, we compared the dose measured with the glass dosimeter and calculated the dose of treatment planning system (TPS) using a cylindrical phantom. Finally, we investigated the feasibility of the glass dosimeter for measurement of dose distributions near the superficial region of 6 patients for proton therapy plans.
Methods and Materials Glass dosimeter system For our study, a model GD-301 glass dosimeter (AGC Techno Glass, Shizuoka, Japan) and FGD-1000 automatic reader were used. Model GD-301 is 1.5 mm in diameter and 8.5 mm in length; the effective readout size is 1 mm in diameter and 0.6 mm in depth (7). Effective atomic number and density of the glass dosimeter were 12.039 and 2.61 g/cm3, respectively. The glass dosimeter was annealed (at 400 C for 1 hour) before it was used for measurement.
Dosimetric characteristics of the proton beam The proton beam reference condition in this study consists of a snout size with an 18 cm diameter and a 10 10 cm2 open block. Uniformity in the response of the glass dosimeters was evaluated for 50 dosimeters with the 200-MeV proton beam in a water phantom. Delivery dose to the dosimeter was approximately 2 Gy at the middle of the SOBP. Dose-linearity evaluates response as a function of the dose delivered to each dosimeter. For determination of their linearity, dosimeters were irradiated with a proton beam in the dose range of 1-10 Gy for the reference condition. For dose-rate dependence, glass dosimeters were irradiated to approximately 2 Gy at each of the following dose rates: 50, 100, 200, 300, 400, and 500 MU/min. For determination of the change in the measured dose with delayed readout time, fading characteristics were investigated under normal ambient conditions during a storage period ranging from 5-60 min postirradiation. Beam profile, which can be obtained from glass dosimeters, was oriented with its axis perpendicular to the beam axis. We irradiated the glass dosimeters with a modulated 160 MeV proton beam. The dosimeters were positioned in a polystyrene plate phantom with a hole. A dose of 2 Gy was delivered to the glass dosimeters at the center of the SOBP. The result measured with the glass dosimeter was evaluated by comparison with those from a plane-parallel Markus type ionization chamber (PTW, TB23343). The diameter and height of the measuring volume are 5 mm and 1 mm, respectively. Depth-dose measurements were 8.0 cm proton range in water, where the mean energy was approximately 160 MeV for both nonmodulated and modulated beams. Measurements were performed in a water phantom using a stairlike holder. The dosimeter was aligned with the center of the beam axis at the isocenter using laser lights. For comparison, depth dose distribution was also measured by the Markus chamber. To determine the varying responses of the glass dosimeter, with proton beam energy, the glass dosimeter was irradiated to a dose of 2 Gy for proton ranges between 5.0 and 19.1 cm. For a given range, a glass dosimeter was placed in a water phantom using a stairlike holder corresponding to the center of the SOBP.
Phantom study Calibration of the glass dosimeter All measurements were performed in a clinical proton beam (IBA Proton Therapy System-Proteus 235) at the National Cancer Center in Korea. The minimum and maximum ranges of the proton beam in patients are 5 cm and 28 cm, respectively, with 0.1 cm accuracy in the double-scattering mode. Glass dosimeters were calibrated by delivery of a known dose in a water phantom using a stairlike holder specially designed for this study. The
The proton dose verification was delivered to an in-house manufactured phantom that was constructed with polymethylmethacrylate, was cylindrical in shape, had an interchangeable plate insert for the dosimeter, was 25 cm in diameter and 20 cm in length. We compared the glass dosimeter and thermoluminescent dosimeter (TLD) dose measurements. The TLDs used in this study were Harshaw TLD-100 lithium fluoride chips with a nominal thickness of 1 mm and a normal surface area of 3 3 mm2.
Volume 84 Number 2 2012
Fig. 1.
In vivo dosimetry with glass dosimeter in proton beams e253
Dose verification measurement using the cylindrical phantom for a proton therapy plan.
The dosimeter evaluation was assessed using 3 separate sets of measurement methods. Dose calculations in Eclipse (Varian Medical Systems, Palo Alto, CA) were performed with a grid size of 2.5 mm. For the first measurement, the superficial planning target volume (PTV) was created by autocontouring of the entire surface of the phantom (Fig. 1a). The surface doses were measured by placing the dosimeters on the front surface of the phantom. The delivered dose for the PTV was 2 Gy with 1 beam angle. In a second set of measurements, the planning target volume (PTV) was localized to the center of the phantom. Dosimeters were then placed at central and surface
points on the phantom, as seen in Fig. 1b. The plan was optimized to give a homogeneous dose of 2 Gy per fraction to the target region with the anterior-posterior beam. Finally, dosimeter measurements were performed in 2 clinical cases with different treatment sites. Figure 1c and 1d shows dose distributions for a proton plan generated by Eclipse. One case was planned for brain cancer with the PTV prescribed to 54 Gy at 2 Gy per fraction. The other case was treated for liver cancer with the PTV prescribed to 55 Gy at 2.5 Gy per fraction. Each experiment was repeated 5 times, and an average measurement was used.
Fig. 2. Dose distribution of treatment proton plans. (a and b) Transverse and sagittal view of patient 1 (breast cancer) without the use of bolus. (c and d) Transverse and sagittal view of patient 3 (inguinal cancer) with the use of bolus.
e254 Rah et al.
International Journal of Radiation Oncology Biology Physics
Because the glass dosimeter and TLD used in this study had a physical thickness of 1.5 mm and 1 mm, respectively, these measurements overestimate the entrance dose. It is important to note that the dosimeters dose results reported in this study are not true surface dose measurements. Therefore, to account for any uncertainty in the positioning of the dosimeters, the positioning of the phantom and the error from the 2.5 mm dose grid, the dosimeter region of interest was extended by 1 voxel outside the visible dosimeters locations. The mean dose and the standard deviation were obtained from calculated dose values bound within the contoured volume.
Patient study Glass dosimeters were used on 6 patients for in vivo dose measurements. We selected patients with clinical localization near the superficial target, which were treated with proton radiation therapy at our institution. The proton therapy cases were planned using the Eclipse for all patients. Figure 2 shows the treatment plan for breast cancer (patient 1) and anal cancer (patient 2), respectively, along with the dose distribution for the corresponding coronal and axial views. Patients 3 and 4 were treated by adding a 10-mm bolus to increase the skin dose. One fraction was delivered and measured with a glass dosimeter and TLD-100 chip for each patient. The dosimeters were arranged into 2 packets, with 3 glass dosimeters and TLD chips per packet on the patient’s skin.
Results and Discussion Dosimetric characteristics of the proton beam Response uniformity was estimated by taking 1 standard deviation of the average for the glass dosimeter. The uniformity result was within 1.5%. Unlike the TLD, which can only be read once, the radiophotoluminescence (RPL) signal does not disappear after the reading operation. Therefore, the glass dosimeter can be read an unlimited number of times (7-10). This provides an advantage in performance of in vivo dosimetry on patients, which requires accurate measurement. The glass dosimeter was found to have a good linear relationship for a dose ranging from 1-10 Gy with a linearity coefficient R of 0.999. The dose-rate dependence was negligible for in vivo purposes and was within 1.5%. The fading effect of the glass dosimeter was relatively stable, within 1.5% for a time delay of 60 min. In our previously published study, the glass dosimeter also presented long-term stability due to less fading. After 150 days of storage, losses in the response of the glass dosimeter did not exceed 2% (8). Figure 3a shows the comparison of beam profiles as measured by the glass dosimeter and Markus chamber in the horizontal plan at the center of the SOBP. It shows the flatness of the proton beam, as measured by the glass dosimeter, to be 1.06 in the horizontal plane. For the ionization chamber, the flatness was 1.02 in the horizontal plane. The beam penumbra between 80% and 20%, obtained from glass dosimeter, was 7.3 mm and was 7.0 mm with the ionization chamber. The profiles from both dosimeters are in agreement at an open block of 10 10 cm2. A comparison of depth-dose distribution for a 160-MeV nonmodulated and modulated proton beam obtained with the glass
Fig. 3. (a) Comparison of beam profile as measured with the glass dosimeter and Markus ionization chamber in horizontal plane. The profiles are normalized at the central axis of a beam. (b and c) Comparison of the Bragg curves obtained with the glass dosimeter and Markus ionization chamber for unmodulated and modulated proton beam. All data are normalized at beam entrance. dosimeters and the ionization chamber is presented in Fig. 3b and 3c. The ratio of the dose measured in the peak to the dose in the entrance region of the Bragg peak was 3.00 with the ionization chamber and 2.93 with the glass dosimeter; it was found to be within 2.5%. This shift was probably due to setup uncertainty in the glass dosimeter positioning of the effective measurement point in a water phantom using a stairlike holder. The range measured with the glass dosimeter was estimated to be 1.9% smaller than
Volume 84 Number 2 2012 Table 1
Set 1 Set 2 Set 3
In vivo dosimetry with glass dosimeter in proton beams e255
Results of the phantom study Measurement position
No. of irradiation fields
Calculation dose (Gy)
Surface Surface Isocenter Case 1: Brain isocenter Case 2 : Liver isocenter
1 1
2.00 1.40 2.00 2.05 2.48
4 3
Glass dosimeter
TLD
Mean (Gy) % SD
Mean (Gy) % SD
1.91 1.29 2.08 1.98 2.41
1.7 1.4 1.2 1.6 1.3
1.84 1.31 1.96 1.91 2.57
2.2 2.9 2.7 2.1 2.3
Abbreviations: SD Z standard deviation; TLD Z thermoluminescent dosimeter.
those from the ionization chamber. The depth-dose distribution of the spread-out Bragg curve obtained with the glass dosimeter was approximately 3% lower than with the ionization chamber at the end of the plateau. However, our results show that the glass dosimeter is more stable and relatively independent of linear energy transfer (LET) than the MOSFET and scintillator dosimeter, especially for large values of dE/dX at lower energies. Kohno et al (3) used the MOSFET for evaluation of characteristics for therapeutic proton beams. The Bragg peak obtained by the MOSFET was estimated to be approximately 40% lower than those by the ionization chamber, demonstrating that MOSFET responses are strongly dependent on LET for proton beams. Torrisi et al (5) evaluated the use of a plastic scintillator dosimeter in a proton beam; however, a plastic scintillator has never been used extensively for proton dose measurements. The main disadvantage of using scintillation detectors in proton therapy is the well-known quenching of the scintillator. Energy dependence of the glass dosimeter is within 3% for proton beam ranging from 5 cm-19.1 cm. Zullo et al (1) reported that energy dependence of Lithium Fluoride TLD powder for proton beams from 100-250 MeV agree within 5% of the actual dose values when the TLD was irradiated in a region of relatively uniform LET values. A recent article by Reft (4) reported on energy dependence of a carbon-doped aluminum oxide (Al2O3:C) as an OSL detector for proton beams. These results demonstrate that OSL response was found to be independent of the proton energy from 150-250 MeV and approximately 6% greater than their response to a 6-MV photon beam. The glass dosimeter is therefore well suited to surface dose measurements. They could be used to measure the entrance dose at the skin precisely.
Phantom study Results of 3 sets of glass dosimeter measurements for plan verification using the cylindrical phantom are summarized in Table 1. Table 2
Each point represents the average value of 5 measurements. Calibration of dosimeters was carried out at the center of the SOBP in a water phantom with a modulated 200 MeV proton beam. In the first and second set of measurements, the difference between the glass dosimeter measured dose and the calculated treatment plan dose at each surface point was 4.7% and 8.5%, respectively. This large deviation was investigated by review of the surface dose calculations performed by the Eclipse TPS (11). Surface dose points were estimated by reading the calculated doses at the intersection of the external contour and air. This boundary was chosen as the depth where the Hounsfield number changes from negative to positive. Since the dose calculation voxels were 2.5 mm3, some averaging takes place. Consequently, surface dose points extracted from the TPS also include uncertainties. The dose assigned to voxels near the air-surface interface was found to be significantly affected by the partial volume effect, leading to inaccurate dose estimation at points on the top of the phantom. On the other hand, for the second and third set, differences of isocenter dose between the calculation from Eclipse and measurement using the glass dosimeter are within approximately 5%. Concurrent TLD measurement differs from calculation by approximately 8%.
Patient study Calculated and measured surface doses delivered to patients are shown in Table 2. The difference between the glass dosimeter measured dose and the calculated treatment plan dose was as large as 11%, with an average of 5.9% 1.8%. Overall average agreement with the TLD and glass dosimeter measurement was within 2%. For surface dose, agreement with the treatment plan doses was similar to the results found with the phantom study. Doses measured by the TLD indicated that the proton therapy Eclipse TPS overestimates surface doses by 3%-16%, except for in patient 4, for whom it underestimated these doses by approximately 2.4%. Doses calculated and measured using the glass
Patient characteristics, prescription dose, and results of the patient study
Glass dosimeter TLD No. of treatment Prescribed Calculation Pt. no. Treatment site Age (y)/sex fields Bolus dose to PTV dose (Gy) Mean (Gy) % SD Mean (Gy) % SD 1 2 3 4 5 6
Breast Anal Inguinal Brain Rib Mediastium
53/F 27/F 54/F 37/M 64/M 71/M
2 2 2 3 3 3
þ þ -
2.00 2.00 2.00 2.20 3.00 2.40
1.89 1.96 1.85 2.07 2.92 2.17
1.79 1.77 1.78 2.00 2.82 2.01
1.9 1.8 1.6 1.9 1.8 1.4
1.82 1.72 1.71 2.12 2.75 1.88
2.8 2.6 2.1 2.7 3.2 2.5
Abbreviations: F Z female; M Z male; PTV Z planning target volume; SD Z standard deviation; TLD Z thermoluminescent dosimeter.
International Journal of Radiation Oncology Biology Physics
e256 Rah et al. Table 3 Uncertainty analysis for measured results obtained with glass dosimeters Physical quantity of procedure Step 1: Calibration procedure Dose measured with the ionization chamber Reading uniformity Setup error Combined uncertainty Step 2: Dose correction factor Combined uncertainty (dose rate, fading, energy dependence, etc) Step 3: In vivo dosimetry Combined uncertainty Overall uncertainty
Uncertainty (%)
measurement. In addition to the uniformity of in vivo dosimeters, components of this step establish the patient setup of conditions for clinical treatment. Therefore, we determined that the uncertainty in surface dose measurement with the glass dosimeter in the proton beam is approximately 4.3%.
2.0 1.2 0.5 2.4 2.9
2.0 4.3%
All uncertainties were estimated as 1 standard deviation.
dosimeter for clinical cases in which the target volume extends to the surface of the patient indicate that the target was underdosed by an average of 7% without the use of bolus. On the other hand, for patients 3 and 4, with use of a bolus to cover the treatment site, calculated values with Eclipse overestimated the measured dose by approximately 3.7%. As such, addition of bolus for these clinical cases is recommended. For patient 2, without the use of bolus, given the result of the measurements using the glass dosimeter and TLD-100, calculated doses on the surface of the patient were overestimated by between 10.7% and 14%. In accordance with our results, the treatment plan for patient 2 was modified for the second fraction to increase the tumor volume coverage of the prescribed dose. The PTV was recontoured including bolus material as a target region. The tumor volume coverage can be achieved with a surface dose to 95% of the prescribed dose. These results confirmed the inaccuracy of the calculation of surface doses by the TPS. Inaccurate calculations in the air-surface interface region will also affect the doses given to the normal skin tissues and near-superficial tumors. Therefore, it is important to measure the dose for near-surface tumors and also to measure the dose delivered to the skin.
Uncertainty analysis Table 3 shows the uncertainty analysis of the glass dosimeter. Evaluation of these uncertainties is important in experimental investigations. The combined standard uncertainty at each step is assessed by the square root of the quadratic sum of the individual uncertainties in dose measurement using the glass dosimeter. For the first step (ie, dose measurement with the reference ionization chamber), typical uncertainties in the clinical proton beam was approximately 2%. For calibration, setup uncertainty was based on the positioning accuracy in the phantom and the accuracy of the source-to-dosimeter distance. This component is also included in daily variations, machine setup, and machine instability during several months of calibration. Uncertainty of the second step could be regarded as the accuracy with which we could determine the correction factors for dosimetric characteristics of the proton beam. The uncertainty of the third step deals with in vivo
Conclusion We evaluated the dosimetric characteristics of the glass dosimeter for in vivo dosimetry in proton beams. Depth-dose distributions in nonmodulated and modulated proton beams obtained with the glass dosimeter were within 3%, lower than those with the ionization chamber. In the phantom study, the difference of isocenter dose between delivery dose calculated by the Eclipse and that measured by the glass dosimeter was within 5% when the dosimeter was calibrated at the center of the SOBP with the modulated proton beam. In vivo dosimetry, the calculated surface doses overestimated measurements by 4%-16% using the glass dosimeter and TLD. It is recommended that a bolus be used on the surface to ensure that tumors of superficial regions are not underdosed. From these results, we conclude that the glass dosimeter has considerable potential as a new dosimeter using a clinical proton beam.
References 1. Zullo JR, Kudchadker RJ, Zhu XR, et al. LiF TLD-100 as a dosimeter in high energy proton beam therapydcan it yield accurate results? Med Dosim 2009;35:63-66. 2. Grusell E, Medin J. General characteristics of the use of silicon diode detectors for clinical dosimetry in proton beams. Phys Med Biol 2000; 45:2573-2582. 3. Kohno R, Nishio T, Miyagishi T, et al. Experimental evaluation of a MOSFET dosimeter for proton dose measurements. Phys Med Biol 2006;51:6077-6086. 4. Reft CS. The energy dependence and dose response of a commercial optically stimulated luminescent detector for kilovoltage photon, megavoltage photon, and electron, proton, and carbon beams. Med Phys 2009;36:1690-1699. 5. Torrisi L. Plastic scintillator investigations for relative dosimetry in proton-therapy. Nucl Instrum Methods Phys Res 2000;170:523-530. 6. Daftari I, Castenadas C, Petti PL, et al. An application of GafChromic MD-55 film for 67.5 MeV clinical proton beam dosimetry. Phys Med Biol 1999;44:2735-2745. 7. Araki F, Ikegami T, Ishidoya T, et al. Measurements of gamma-knife helmet output factors using a radiophotoluminescent glass rod dosimeter and a diode detector. Med Phys 2003;30:1976-1981. 8. Rah J, Hong J, Kim G, et al. A comparison of the dosimetric characteristics of a glass rod dosimeter and a thermoluminescent dosimeter for mailed dosimeter. Radiat Meas 2009;44:18-22. 9. Rah J, Hwang U, Jeong H, et al. Clinical application of glass dosimeter for in vivo dose measurements of total body irradiation treatment technique. Radiat Meas 2011;46:40-45. 10. Absorbed Determination in External Beam Radiotherapy: an International Code of Practice for Dosimetry Based on Standard Absorbed Dose to Water [Technical Report Series No. 398]. Vienna: International Atomic Energy Agency; 2000. 11. Tournel K, Verellen D, Duchateau, et al. An assessment of the use of skin flashes in helical tomotherapy using phantom in-vivo dosimetry. Radiother Oncol 2007;84:34-39.