doi: /j.ijrobp

doi:10.1016/j.ijrobp.2005.05.021 Int. J. Radiation Oncology Biol. Phys., Vol. 63, No. 2, pp. 577 583, 2005 Copyright 2005 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/05/$ see front matter PHYSICS CONTRIBUTION DESIGN AND IMPLEMENTATION OF AN ANTHROPOMORPHIC QUALITY ASSURANCE PHANTOM FOR INTENSITY-MODULATED RADIATION THERAPY FOR THE RADIATION THERAPY ONCOLOGY GROUP ANDREA MOLINEU, M.S.,* DAVID S. FOLLOWILL, PH.D.,* PETER A. BALTER, PH.D.,* WILLIAM F. HANSON, PH.D.,* MICHAEL T. GILLIN, PH.D.,* M. SAIFUL HUQ, PH.D., AVRAHAM EISBRUCH, M.D., AND GEOFFREY S. IBBOTT, PH.D.* *Department of Radiation Physics, The University of Texas M. D. Anderson Cancer Center, Houston, TX; Department of Radiation Oncology, University of Pittsburgh Medical Center, Pittsburgh, PA; Department of Radiation Oncology, University of Michigan Medical Center, Ann Arbor, MI Purpose: To design, construct, and evaluate an anthropomorphic phantom for evaluation of intensity-modulated radiation therapy (IMRT) dose planning and delivery, for protocols developed by the Radiation Therapy Oncology Group (RTOG) and other cooperative groups. Methods and Materials: The phantom was constructed from a plastic head-shaped shell and water-equivalent plastics. Internal structures mimic planning target volumes and an organ at risk. Thermoluminescent dosimeters (TLDs) and radiochromic film were used to measure the absolute dose and the dose distribution, respectively. The reproducibility of the phantom s dosimeters was verified for IMRT treatments, and the phantom was then imaged, planned, and irradiated by 10 RTOG institutions. Results: The TLD results from three identical irradiations showed a percent standard deviation of less than 1.6%, and the film-scanning system was reproducible to within 0.35 mm. Data collected from irradiations at 10 institutions showed that the TLD agreed with institutions doses to within 5% standard deviation in the planning target volumes and 13% standard deviation in the organ at risk. Shifts as large as 8 mm between the treatment plan and delivery were detected with the film. Conclusions: An anthropomorphic phantom using TLD and radiochromic film can verify dose delivery and field placement for IMRT treatments. 2005 Elsevier Inc. Intensity-modulated radiation therapy, Quality assurance, Anthropomorphic phantom, Cooperative groups. INTRODUCTION Intensity-modulated radiation therapy (IMRT) has gained acceptance as an improved treatment technique for several disease sites (1 3). Several manufacturers of radiation therapy equipment provide devices to enable the delivery of IMRT, including multileaf collimators and inverse-planning (optimization) treatment-planning systems. Because IMRT offers the possibility of high dose gradients, it is possible to deliver high doses to target volumes while maintaining low doses to nearby critical normal structures to a much greater extent than is the case with conventional radiation therapy. At the same time, the high dose gradients achievable with IMRT mean that localization of the dose distribution is critical. Small errors in positioning of the patient can mean that a target volume is missed or that a sensitive normal Reprint requests to: Andrea Molineu, M.S., Department of Radiation Physics, The University of Texas M. D. Anderson Cancer Center, Unit 547, 1515 Holcombe Blvd., Houston, TX 77030. Tel: (713) 745-8989; Fax: (713) 794-1364; E-mail: amolineu@ mdanderson.org This work was supported by Public Health Service Grant CA 81647, awarded by the National Cancer Institute, Department of structure is irradiated to a higher dose than intended, and perhaps higher than can be tolerated. Consequently, comprehensive quality assurance (QA) procedures are necessary (4). A QA program for IMRT should focus on the characteristics that strain the abilities of the procedures and equipment. For example, the nature and advantages of IMRT demand that patient positioning be considerably more precise and reproducible than for conventional treatment. Devices to facilitate reproducible positioning and patient immobilization are available for imaging, simulation, and treatment equipment, and QA procedures should be implemented to ensure proper function and correct usage. Other alignment devices, such as lasers, video systems, and ultrasound imaging systems, should be evaluated on a regular basis to ensure their correct function. Health and Human Services. Acknowledgments The authors thank the Radiation Therapy Oncology Group for its contributions, in particular Drs. Robert Kline and James Galvin for their helpful discussions. Received Jan 30, 2004, and in revised form May 5, 2005. Accepted for publication May 11, 2005. 577

578 I. J. Radiation Oncology Biology Physics Volume 63, Number 2, 2005 Before a patient is treated with IMRT, it is the standard of practice to perform some form of plan verification. This might have two or more components, generally including the validation of data transfer and a dosimetric confirmation of the intended dose delivery. Confirming delivery of the intended dose requires that the treatment be delivered with some sort of phantom containing dosimeters in place of the patient. Inverse planning systems allow the planned intensity distribution or leaf sequences to be applied to a convenient geometrically shaped (generally elliptical or cylindrical) phantom. Dosimeters (such as ionization chambers or thermoluminescent dosimeters [TLDs]) are then placed at selected locations in the phantom, and the complete IMRT sequence is delivered. Delivery of the expected dose at the locations selected for measurement is taken as an indication that the correct dose distribution is being delivered. The Radiologic Physics Center (RPC) (http://rpc. mdanderson.org/rpc/) is funded by the National Cancer Institute to assure the cooperative study groups that conduct multi-institutional clinical trials that institutions participating in the trials have adequate QA procedures and that no major systematic dosimetry discrepancies exist. The RPC monitors the linear accelerator output at more than 1300 institutions and verifies dosimetry data used by the institution, brachytherapy source strength, the calculation algorithms used for treatment planning, and the institution s QA procedures. Remote monitoring procedures include the use of mailed TLDs to verify machine output, comparison of an institution s dosimetry data with RPC standard data to identify potential discrepancies, evaluation of reference or actual patient calculations to verify the treatment-planning algorithms and manual calculations, review of the institution s written QA procedures and records to verify adherence to Task Group 40 (TG-40), and mailed anthropomorphic phantoms to verify tumor dose delivery for special treatment techniques. The RPC is a member of the Advanced Technology Consortium (http://atc.wustl.edu; member organizations are listed on the Members page), which is funded by the National Cancer Institute to provide quality assurance for advanced technology in radiation therapy clinical trials for protocol groups in the United States. Among other things, the Advanced Technology Consortium enables electronic submission of treatment data and conducts evaluations of protocol compliance and credentialing through the RPC. Recently, the Radiation Therapy Oncology Group (RTOG) and other cooperative groups have begun to evaluate the use of IMRT for treatment of patients submitted to some clinical trials, and other study groups have indicated their intention to begin such trials in the near future. Study groups are interested in ensuring that their participating institutions practice good QA so that patients treated with IMRT are treated safely, effectively, and in compliance with the protocol. To enter patients treated with IMRT into a clinical trial, institutions are required to go through a credentialing process that demonstrates that they meet the requirements of the study group. The RPC currently participates in the credentialing of institutions wishing to enter patients into clinical trials involving IMRT. The RPC provides anthropomorphic dosimetry QA phantoms to verify an institution s ability to deliver IMRT treatments appropriately. This report describes the development and initial evaluation of an anthropomorphic phantom for credentialing purposes for RTOG IMRT protocols. METHODS AND MATERIALS Phantom design The IMRT head-and-neck phantom was designed to meet certain design criteria, including having an anthropomorphic outer plastic shell so that it tests realistic anatomic clinical situations, being lightweight so that it is inexpensive to mail, using water as a substitute for tissue where possible, and containing a target that includes radiation dosimeters and that can be imaged with computed tomography (CT). The outer plastic shell of the head was purchased (The Phantom Laboratory, Salem, NY) and modified in the M. D. Anderson Cancer Center Department of Radiation Physics machine shop to hold the imaging/dosimetry insert in a watertight environment. All air spaces surrounding the imaging/ dosimetry insert could be filled with water, and all modifications used plastic so that no metal parts would interfere with the imaging of the phantom. The exterior plastic shell was of sufficiently realistic anthropomorphic shape that the head phantom would fit in most treatment immobilization devices used for IMRT treatments. Figure 1 shows the hollow phantom head with the insert in place. The crown of the head is removable so that the insert can be placed into its watertight housing. The base contains a port that allows the shell to be filled with water and is fitted with adjustable screws so that the head can be placed on the type of headrest commonly used by the institution. The insert was constructed as a block of polystyrene housing solid water targets and an acrylic organ at risk (OAR). The planning targets were designed in collaboration with the RTOG Medical Physics Committee to mimic an RTOG oropharyngeal protocol (H-0022) that would have primary and secondary targets with an OAR adjacent to the primary target. The solid water used for the targets and acrylic used for the OAR were of slightly Fig. 1. The intenstity-modulated radiation therapy head-and-neck phantom, with insert in place.

Anthropomorphic QA phantom for IMRT A. MOLINEU et al. 579 Fig. 2. The dosimetry/imaging insert. The primary planning target volume (PTV), secondary PTV, organ at risk (OAR), and holes for the thermoluminescent dosimeter (TLD) capsules can be seen. The axial film is placed in the cross-section shown, and the sagittal film pieces are placed in the slits though the primary PTV. different densities than the surrounding water and plastics for imaging and target identification purposes. However, the differences in density and atomic number were small enough to have an insignificant effect on treatment delivery. The insert was designed to hold TLDs in the two planning target volumes (PTVs) and the OAR. Radiochromic films (RCFs) were placed in the axial and sagittal planes through the primary target and in the axial plane through the secondary target and OAR. Figure 2 shows the 7.5 cm 10.5 cm 13 cm polystyrene insert. The primary and secondary PTVs are made of solid water and are 5 cm long. The primary PTV is 4 cm in diameter and holds two TLDs. One TLD is located superior to and the other inferior to the axial film. The secondary PTV is 2 cm in diameter and has one TLD in its center, directly inferior to the axial film. The centers of the two PTVs are 5.2 cm apart. The OAR is made of acrylic, is 1 cm in diameter, and extends the length of the insert. It has one TLD in the center of its cross section, located directly inferior to the axial film. The edge of the OAR is 0.8 cm from the edge of the primary PTV in the posterior direction. The locations of the RCFs can also be seen in Fig. 2. A single sheet is placed in the axial plane. The slits through the primary PTV and OAR indicate the placement of the two sagittal films inferior and superior to the axial film. Figure 3 is an axial CT slice of the phantom. Both PTVs, the OAR, and a film slit are visible. Because the dosimeters would be irradiated during both the imaging and the treatment delivery, TLDs were located on the exterior of the phantom at the location of the ears during the imaging irradiation. These TLDs were used to subtract the dose given to the dosimeters in the phantom as a result of the imaging process from the dose given during treatment. The imaging dose was typically on the order of 1% of the dose given to the PTV. Phantom dosimeters TLD-100 powder, placed in custom-built, reusable capsules, is used as the absolute dosimeter. The capsules are small cylinders with outer dimensions of 5 mm (height) 5 mm (diameter), with a 1-mm wall thickness. Each capsule holds approximately 40 mg of powder that yields two readings. The capsules are constructed of high-impact polystyrene that contains very little air and approximate a small sphere. The capsules are irradiated by the remote institution and are read at the RPC with the same technique as is used in the TLD mail-out dosimetry program. The TLD-100 powder is taken from the TLD powder stock used by the RPC for its mail-out dosimetry service (5). Each batch of TLD-100 powder used in the mail-out system is evaluated by the RPC for dose response, energy dependence, dose uniformity, and fading. It has been shown that the accuracy of absolute dose determined in the TLD readout system is equivalent to ion chambers to within 4% at a 90% confidence interval (6). The TLD system is also precise to within 3% and is capable of detecting dose errors on the order of 5% or greater (5). Thermoluminescent dosimeter dose readings are used to evaluate the absolute dose delivered to the phantom, and the two capsules on either side of the target center are used to normalize the film dose distributions. The exceptions to the published TLD reading procedure are that a filter is used to damp the thermoluminescent signal, and high-dose TLD standards are used because of the high doses delivered to the phantom. Supralinearity of the thermoluminescent signal is accounted for in the dose calculations. Radiochromic films (GAFChromic MD-55-2 film; Nuclear Associates, Carle Place, NY) were used to measure dose distributions and field localization. Radiochromic film is approximately tissue equivalent and has no significant angular dependence (7). It is relatively insensitive to room light and thus easy to work with in the mail-out program. Radiochromic film was investigated for dose response, fading, energy independence, and uniformity and was Fig. 3. Computed tomographic slice of phantom.

580 I. J. Radiation Oncology Biology Physics Volume 63, Number 2, 2005 Table 1. Reproducibility results for the TLD doses averaged for three irradiations TLD Average (cgy) SD Primary PTV (superior) 714 8.1 Primary PTV (inferior) 804 10.8 Secondary PTV 663 6.7 Organ at risk 178 2.8 Abbreviations: TLD thermoluminescent dosimeter; SD standard deviation. found to have high spatial resolution and low spectral sensitivity, making it well suited to measuring IMRT-produced high dosegradient radiation fields (8, 9). Two sheets of RCF were placed in the insert in the axial and sagittal planes. The sagittal film was cut into two pieces, allowing the axial film to intersect it at the center of the primary target. Small holes were drilled in the dosimetry insert to allow two localization marks to be made on each piece of film. These marks uniquely specify the location and orientation of the films with respect to the targets. The recommendations for the handling and evaluation of RCF for use in dosimetry were followed (7). Films were scanned with a 633-nm laser densitometer (Personal Densitometer; Molecular Dynamics, Sunnyvale, CA) with a scanning resolution of 0.1 mm. Films irradiated in the phantom were scanned on a ground glass scanning bed to eliminate interference-pattern artifacts found with use of the standard polished glass scanning bed. A mask was created for the scanning bed to exclude light contamination. This film-scanning system is based on the technique used by Dempsey et al. (10). Films were scanned at least 36 hours after irradiation to avoid documented reverse-fading effects (7). All films from the same lot were stored together so that they had the same temperature, light, and non treatment-related radiation exposure history. A dose response curve from 0 Gy to 30 Gy was measured for each lot of film used in this analysis. The dose response curve was used to correct for any nonlinearity in the dose response of the film. The variations in film sensitivity were previously measured at the RPC by irradiating multiple sheets of film from the same lot to determine reproducibility of the film response. A background film from the same lot and order was included and scanned with all treatment films. A 12-bit TIFF (tagged image file format) image file was generated from the film scan in an array of optical density (OD) values ranging from 0 to 4095. The background OD values were subtracted from each irradiated film. The average background OD was 0.24. Profiles through the center of the primary PTVs and OARs were generated according to localization marks on each film. These profiles were normalized to the TLD results. Pixel averaging was used to smooth the profiles (11). Reproducibility studies The phantom was imaged at M. D. Anderson Cancer Center in an ACQSIM CT scanner (Philips Medical Systems, Cleveland, OH), and a benchmark treatment plan was generated with the Corvus planning system (NOMOS, Sewickley, PA). The dose prescription, which was a factor of 10 lower than the RTOG protocol, was 6.6 Gy to at least 95% of the primary PTV and 5.4 Gy to at least 95% of the secondary PTV. Only 1% or less of the primary and secondary PTVs could receive less than 93% of the prescription dose. The OAR was to receive less than 4.5 Gy, and the normal tissue could not receive more than 110% of the primary PTV prescription dose. The phantom was placed on the treatment couch and positioned with lasers and phantom fiducials. The dose was delivered with a nine-field step-and-shoot arrangement with the 6-MV X-ray beam of a Varian Clinac 2100CD (Varian Medical Systems, Palo Alto, CA). Once each treatment was delivered, the dosimeters were removed, and new ones were placed in the phantom without disturbing the phantom position on the treatment couch. The phantom was then re-irradiated. A total of three phantom irradiations were delivered. Data comparison The phantom was mailed to 10 institutions wishing to participate in an RTOG IMRT protocol. Each institution was instructed to image the phantom, design a treatment plan in accordance with the prescription outlined above, and treat the phantom according to the plan. Each institution irradiating the IMRT head-and-neck phantom was instructed to provide dose calculations and dose distribution information for comparison with the TLD and RCF measurements. The institution was asked to outline the TLD powder and report the minimum, mean, and maximum dose to the TLD capsule. The institution was also asked to provide dose distributions in the planes corresponding to the location of the films. The treatment plan was compared with the measured dose profiles and TLD results in the two planes of measurement. Ratios of the RPC s measured TLD dose to the institution s reported mean dose to the TLD powder were determined for the TLDs in the primary and secondary PTVs. Film profiles through the center of the primary PTV were scaled to the TLD dose values. The subsequent film profile results were then plotted against the dose profile data taken from the institution s treatment-planning computer-generated isodose lines. The distance between the measured dose gradient and the institution s calculated dose gradient was determined in the region between the OAR and the primary PTV. It was measured at three levels (25%, 50%, and 75% of the difference between the maximum and minimum measured doses in the region between the OAR and the primary PTV) and averaged. RESULTS Phantom dosimetry reproducibility To determine how well the TLD results could be reproduced, a benchmark treatment plan for the phan- Fig. 4. A plot of a right left profile comparing three irradiations of the phantom done in the same evening with the predicted treatment planning values (black circles). PTV planning target volume.

Anthropomorphic QA phantom for IMRT A. MOLINEU et al. 581 Table 2. Ratio of the RPC TLD dose to the institution s reported dose and dose profile displacement analysis for the 10 institutions that irradiated the phantom Institution Primary PTV (superior) Primary PTV (inferior) Secondary PTV Profile displacement (mm) Organ at risk 1 0.97 1.04 1.10 1.3 0.99 2 0.93 0.96 0.93 2.0 0.82 3 1.04 1.01 1.0 1.16 4 0.96 1.00 0.98 0.1 5 1.09 1.07 1.11 3.3 1.35 6 1.01 1.03 0.99 3.7 1.10 7 1.02 1.03 1.00 2.3 0.95 8 0.98 0.98 0.97 0.1 0.88 9 1.04 1.04 1.06 0.7 1.09 10 1.04 1.05 1.04 4.0 1.18 Average 1.01 1.02 1.02 1.8 1.07 % SD 4.8 3.5 5.7 1.5 15.6 Abbreviations: RPC Radiologic Physics Center; PTV planning target volume. Other abbreviations as in Table 1. tom was designed. The plan was delivered three times in one evening. There was minimal disturbance to the phantom setup between irradiations. The combined TLD results from these irradiations are shown in Table 1. The percent standard deviation was 1.6% for each of the four measurement points. A representative film pro- Fig. 5. A plot of (a) a superior inferior profile taken from a sagittal film; (b) a right left profile taken from an axial film; and (c) an anterior posterior profile taken from an axial film. The black points are the institution data taken from the treatment plan, and the gray line is the film data. The film data closely agree with the institution plan on all three profiles. PTV planning target volume.

582 I. J. Radiation Oncology Biology Physics Volume 63, Number 2, 2005 Fig. 6. A plot of (a) a superior inferior profile taken from a sagittal film; (b) a right left profile taken from an axial film; and (c) an anterior posterior profile taken from an axial film. The black points are the institution data taken from the treatment plan, and the gray line is the film data. The superior inferior film data show that the phantom was shifted in the inferior direction between planning and treatment. The anterior posterior graph includes the regression lines used to find displacement. The levels of measurement are marked with arrows. There is an average displacement of 2 mm in the anterior posterior direction. PTV planning target volume. file that compares the three irradiations is shown in Fig. 4. To determine the reproducibility of our film-scanning system, the films from one irradiation were scanned three times. For each scan, profiles in the two planes were normalized to unity at the center of the primary PTV. The profiles from the three scans were compared in a highgradient region at a level of 75%. The maximum uncertainty in film reproducibility for the three scans was 0.35 mm. Phantom irradiation Ten institutions irradiated a phantom as part of an initial evaluation program. The TLD dose results from these irradiations are shown in Table 2. The results are presented as the ratio of the RPC TLD reading to the average dose to TLD as predicted by the institution. Table 2 also shows the displacement results in the region between the OAR and the primary PTV for the 10 institutions. Two of the TLD results are missing because the primary PTV inferior TLD for Institution 3 was lost, and Institution 4 was not able to give a mean dose to the OAR TLD capsule. There was agreement between the doses measured by TLD and those reported by the institutions in the PTVs, with the largest standard deviation being 5.7% for the results in the secondary PTV. The dose agreement in the OAR was not as close, with an average TLD/institution ratio of 1.07 and a standard deviation of 15.6%. The OAR TLD results correlate fairly well with the profile displacement results. The dose gradient in the region between the OAR and the primary PTV is very steep, and any small displacement will result in a large difference in dose delivery to the OAR. In all cases, the RPC TLD dose fell between the minimum and maximum doses reported by each institution for each measurement point. Gross TLD differences in the PTVs and OARs can indicate errors in positioning the phantom before irradiation.

Anthropomorphic QA phantom for IMRT A. MOLINEU et al. 583 The institutions were asked to treat the phantom as if it were a real patient and to pay special attention to setup accuracy. An error of more than 200% was seen in the OAR TLD for one institution. Visual inspection of the RCF suggested that the phantom had been shifted in the anterior direction relative to the plan, resulting in the large dose difference seen with the TLD. This mistake was confirmed by the institution. Further analysis of errors in IMRT treatment-planning systems that were confirmed by phantom irradiations was done by Cadman et al. (12). Cadman et al. indicated in their article that improper modeling of rounded multileaf collimator leaf ends might result in significant dose discrepancies. The results in Table 2 were used to develop TLD and profile displacement criteria in collaboration with RTOG. Institution 5 was not included in the data analysis to develop the acceptance criteria because it reported a point dose to the TLD, not an average dose. Institution 5 was unable to retrieve the necessary data owing to an upgrade in treatment-planning software at the institution between the time of phantom irradiation and TLD evaluation. The evaluation criteria were set at 7% for the primary and secondary PTVs. This value was 1.64 times the standard deviation, excluding Institution 5. The profile displacement criterion for the dose gradient between the OAR and the primary PTV was set at 4 mm, which is the range of average displacements, once again excluding Institution 5. Film profiles from the axial film were taken through the center of the primary PTV in the anterior posterior and right left directions. A profile in the superior inferior direction was taken from the sagittal film. The profiles were normalized to the TLD results from the primary PTV. Criteria have not yet been determined for the right left and superior inferior profiles, so these films were analyzed qualitatively. Figure 5 shows three film profiles from an institution with fairly good film-to-plan agreement. A significant dose gradient was observed in the OAR region. The size of the TLD is large compared with the gradient. A small shift in position can cause large differences in the dose in the OAR region. The profile shown in Fig. 6 is an example of poor filmto-institution agreement. The film results show that the phantom was shifted 8 mm in the inferior direction relative to the treatment plan. The institution that delivered this treatment has agreed to irradiate the phantom again. DISCUSSION A lightweight anthropomorphic phantom was developed to test dose delivery for IMRT head-and-neck treatments. Thermoluminescent dosimeters and RCF were used as dosimeters. Criteria for the phantom results were agreed upon by the Advanced Technology Consortium and RTOG. The TLDs in the primary and secondary PTVs have a 7% acceptance criterion. Also, the agreement between film measurements and the institution data in the dose gradient region between the primary PTV and OAR must be at least 4 mm. The TLD can detect error on the order of 5%. Because the TLD is larger than dose gradients typically seen in IMRT treatment plans, the size will be modified in the future. Because of the high dose gradients, one-dimensional film profiles do not give a complete dose analysis. A twodimensional comparison would provide a more comprehensive method of determining how the treatment delivered corresponded with the treatment planned. REFERENCES 1. Lee N, Xia P, Fischbein NJ, et al. Intensity-modulated radiation therapy for head-and-neck cancer: The UCSF experience focusing on target volume delineation. Int J Radiat Oncol Biol Phys 2003;57:49 60. 2. Lin A, Kim HM, Terrell JE, et al. Quality of life after parotid-sparing IMRT for head-and-neck cancer: A prospective longitudinal study. Int J Radiat Oncol Biol Phys 2003; 57:61 70. 3. Zhou J, Fei D, Wu Q. Potential of intensity-modulated radiotherapy to escalate doses to head-and-neck cancers: What is the maximal dose? Int J Radiat Oncol Biol Phys 2003;57:673 682. 4. Boyer AL, Mok E, Luxton G, et al. Quality assurance for treatment planning dose delivery by 3DRTP and IMRT. In: Shiu AS, Mellenberg DE, editors. General practice of radiation oncology physics in the 21st century. Madison, WI: Medical Physics Publishing; 2000. p. 187 230. 5. Kirby TH, Hanson WF, Gastorff RJ, et al. System for photon and electron therapy beams. Int J Radiat Oncol Biol Phys 1986;12:261 265. 6. Kirby TH, Hanson WF, Gastorff RJ, et al. Uncertainty analysis of absorbed dose calculations for thermoluminescence dosimeters. Med Phys 1992;19:1427 1433. 7. American Association of Physicists in Medicine Radiation Therapy Committee Task Group 55. Radiochromic film dosimetry: Recommendations of AAPM Radiation Therapy Committee Task Group 55. Med Phys 1998;25:2093 2115. 8. Balter P, Stovall M, Hanson WF. An anthropomorphic head phantom for remote monitoring of stereotactic radiosurgery at multiple institutions. Med Phys 1999;26:1164. 9. Radford DA, Followill DS, Hanson WF. A standard method of quality assurance for intensity modulated radiation therapy of the prostate. Med Phys 2001;28:1211. 10. Dempsey JF, Low DA, Kirov AS, et al. Quantitative optical densitometry with scanning-laser film digitizers. Med Phys 1999;26:1721 1731. 11. Dempsey JF, Low DA, Mutic S, et al. Validation of a precision radiochromic film dosimetry system for quantitative twodimensional imaging for acute exposure dose distributions. Med Phys 2000;27:2462 2475. 12. Cadman P, Bassalow R, Sidhu NPS, et al. Dosimetric considerations for validation of a sequential IMRT process with a commercial treatment planning system. Phys Med Biol 2002; 47:3001 3010.