Search

Purpose: As clinics begin to use 3D metrics for intensity-modulated radiation therapy (IMRT) quality assurance, it must be noted that these metrics will often produce results different from those produced by their 2D counterparts. 3D and 2D gamma analyses would be expected to produce different values, in part because of the different search space available. In the present investigation, the authors compared the results of 2D and 3D gamma analysis (where both datasets were generated in the same manner) for clinical treatment plans. Methods: Fifty IMRT plans were selected from the authors’ clinical database, and recalculated using Monte Carlo. Treatment planning system-calculated (“evaluated dose distributions”) and Monte Carlo-recalculated (“reference dose distributions”) dose distributions were compared using 2D and 3D gamma analysis. This analysis was performed using a variety of dose-difference (5%, 3%, 2%, and 1%) and distance-to-agreement (5, 3, 2, and 1 mm) acceptance criteria, low-dose thresholds (5%, 10%, and 15% of the prescription dose), and data grid sizes (1.0, 1.5, and 3.0 mm). Each comparison was evaluated to determine the average 2D and 3D gamma, lower 95th percentile gamma value, and percentage of pixels passing gamma. Results: The average gamma, lower 95th percentile gamma value, and percentage of passing pixels for each acceptance criterion demonstrated better agreement for 3D than for 2D analysis for every plan comparison. The average difference in the percentage of passing pixels between the 2D and 3D analyses with no low-dose threshold ranged from 0.9% to 2.1%. Similarly, using a low-dose threshold resulted in a difference between the mean 2D and 3D results, ranging from 0.8% to 1.5%. The authors observed no appreciable differences in gamma with changes in the data density (constant difference: 0.8% for 2D vs 3D). Conclusions: The authors found that 3D gamma analysis resulted in up to 2.9% more pixels passing than 2D analysis. It must be noted that clinical 2D versus 3D datasets may have additional differences—for example, if 2D measurements are made with a different dosimeter than 3D measurements. Factors such as inherent dosimeter differences may be an important additional consideration to the extra dimension of available data that was evaluated in this study.

Purpose: For nonsmall cell lung cancer (NSCLC) patients, quantitative image features extracted from computed tomography (CT) images can be used to improve tumor diagnosis, staging, and response assessment. For these findings to be clinically applied, image features need to have high intra and intermachine reproducibility. The objective of this study is to identify CT image features that are reproducible, nonredundant, and informative across multiple machines. Methods: Noncontrast-enhanced, test-retest CT image pairs were obtained from 56 NSCLC patients imaged on three CT machines from two institutions. Two machines (“M1” and “M2”) used cine 4D-CT and one machine (“M3”) used breath-hold helical 3D-CT. Gross tumor volumes (GTVs) were semiautonomously segmented then pruned by removing voxels with CT numbers less than a prescribed Hounsfield unit (HU) cutoff. Three hundred and twenty eight quantitative image features were extracted from each pruned GTV based on its geometry, intensity histogram, absolute gradient image, co-occurrence matrix, and run-length matrix. For each machine, features with concordance correlation coefficient values greater than 0.90 were considered reproducible. The Dice similarity coefficient (DSC) and the Jaccard index (JI) were used to quantify reproducible feature set agreement between machines. Multimachine reproducible feature sets were created by taking the intersection of individual machine reproducible feature sets. Redundant features were removed through hierarchical clustering based on the average correlation between features across multiple machines. Results: For all image types, GTV pruning was found to negatively affect reproducibility (reported results use no HU cutoff). The reproducible feature percentage was highest for average images (M1 = 90.5%, M2 = 94.5%, M1∩M2 = 86.3%), intermediate for end-exhale images (M1 = 75.0%, M2 = 71.0%, M1∩M2 = 52.1%), and lowest for breath-hold images (M3 = 61.0%). Between M1 and M2, the reproducible feature sets generated from end-exhale images were relatively machine-sensitive (DSC = 0.71, JI = 0.55), and the reproducible feature sets generated from average images were relatively machine-insensitive (DSC = 0.90, JI = 0.87). Histograms of feature pair correlation distances indicated that feature redundancy was machine-sensitive and image type sensitive. After hierarchical clustering, 38 features, 28 features, and 33 features were found to be reproducible and nonredundant for M1∩M2 (average images), M1∩M2 (end-exhale images), and M3, respectively. When blinded to the presence of test-retest images, hierarchical clustering showed that the selected features were informative by correctly pairing 55 out of 56 test-retest images using only their reproducible, nonredundant feature set values. Conclusions: Image feature reproducibility and redundancy depended on both the CT machine and the CT image type. For each image type, the authors found a set of cross-machine reproducible, nonredundant, and informative image features that would be useful for future image-based models. Compared to end-exhale 4D-CT and breath-hold 3D-CT, average 4D-CT derived image features showed superior multimachine reproducibility and are the best candidates for clinical correlation.

Purpose: Several simplifications used in clinical implementations of the convolution/superposition (C/S) method, specifically, density scaling of water kernels for heterogeneous media and use of a single polyenergetic kernel, lead to dose calculation inaccuracies. Although these weaknesses of the C/S method are known, it is not well known which of these simplifications has the largest effect on dose calculation accuracy in clinical situations. The purpose of this study was to generate and characterize high-resolution, polyenergetic, and material-specific energy deposition kernels (EDKs), as well as to investigate the dosimetric impact of implementing spatially variant polyenergetic and material-specific kernels in a collapsed cone C/S algorithm. Methods: High-resolution, monoenergetic water EDKs and various material-specific EDKs were simulated using the EGSnrc Monte Carlo code. Polyenergetic kernels, reflecting the primary spectrum of a clinical 6 MV photon beam at different locations in a water phantom, were calculated for different depths, field sizes, and off-axis distances. To investigate the dosimetric impact of implementing spatially variant polyenergetic kernels, depth dose curves in water were calculated using two different implementations of the collapsed cone C/S method. The first method uses a single polyenergetic kernel, while the second method fully takes into account spectral changes in the convolution calculation. To investigate the dosimetric impact of implementing material-specific kernels, depth dose curves were calculated for a simplified titanium implant geometry using both a traditional C/S implementation that performs density scaling of water kernels and a novel implementation using material-specific kernels. Results: For our high-resolution kernels, we found good agreement with the Mackie et al. kernels, with some differences near the interaction site for low photon energies (

Purpose: The aim of this work was to create a mailable phantom with measurement accuracy suitable for Radiological Physics Center (RPC) audits of high dose-rate (HDR) brachytherapy sources at institutions participating in National Cancer Institute-funded cooperative clinical trials. Optically stimulated luminescence dosimeters (OSLDs) were chosen as the dosimeter to be used with the phantom. Methods: The authors designed and built an 8 × 8 × 10 cm3 prototype phantom that had two slots capable of holding Al2O3:C OSLDs (nanoDots; Landauer, Glenwood, IL) and a single channel capable of accepting all 192Ir HDR brachytherapy sources in current clinical use in the United States. The authors irradiated the phantom with Nucletron and Varian 192Ir HDR sources in order to determine correction factors for linearity with dose and the combined effects of irradiation energy and phantom characteristics. The phantom was then sent to eight institutions which volunteered to perform trial remote audits. Results: The linearity correction factor was kL = (−9.43 × 10−5 × dose) + 1.009, where dose is in cGy, which differed from that determined by the RPC for the same batch of dosimeters using 60Co irradiation. Separate block correction factors were determined for current versions of both Nucletron and Varian 192Ir HDR sources and these vendor-specific correction factors differed by almost 2.6%. For the Nucletron source, the correction factor was 1.026 [95% confidence interval (CI) = 1.023–1.028], and for the Varian source, it was 1.000 (95% CI = 0.995–1.005). Variations in lateral source positioning up to 0.8 mm and distal/proximal source positioning up to 10 mm had minimal effect on dose measurement accuracy. The overall dose measurement uncertainty of the system was estimated to be 2.4% and 2.5% for the Nucletron and Varian sources, respectively (95% CI). This uncertainty was sufficient to establish a ±5% acceptance criterion for source strength audits under a formal RPC audit program. Trial audits of four Nucletron sources and four Varian sources revealed an average RPC-to-institution dose ratio of 1.000 (standard deviation = 0.011). Conclusions: The authors have created an OSLD-based 192Ir HDR brachytherapy source remote audit tool which offers sufficient dose measurement accuracy to allow the RPC to establish a remote audit program with a ±5% acceptance criterion. The feasibility of the system has been demonstrated with eight trial audits to date.

Purpose: Failure modes and effects analysis (FMEA) as defined in TG‐100 has become a popular concept throughout the radiotherapy community. This risk mitigation technique involves detailed process mapping, analysis, and ranking of potential errors by means of a subjective, ordinal scoring system. This study aims to reduce the subjectivity of FMEA severity scoring for IMRT delivery by providing quantitative values. Methods: First we created an IMRT delivery process map for physics‐applicable processes and identified 11 physical failure modes (FMs). To determine the magnitude of dose delivery errors for several of the physical FMs (i.e., the severity of the FM), FMs were induced and dosimetry measurements were performed on a Varian Clinac accelerator going out of clinical service. Treatment planning studies to simulate remaining FMs are to follow. The quantitative severity scores will be compared to recommended FMEA subjective scores. Results: We identified the following physical FMs to investigate: photon beam energy, symmetry, MLC position, MLC transmission and leakage, MLC rounded‐end and leaf offset, MLC tongue‐and‐groove, CT look‐up table, gantry angle, collimator angle, couch angle and displacement, and MU linearity. Within the PTV of a H&N IMRT phantom, the maximum dose delivery error of 3% absolute dose was seen for 2 mm systematic MLC offsets, 4% for 1.1% increased energy, and 4% for 3.5% symmetry error. Measurements will be compared to computational study results and then used for quantitative severity scoring determination. Conclusion: Current FMEA practice for radiotherapy requires quantitative data in order to make accurate assessments associated with clinical QA programs. This study has shown examples of error magnitudes induced by IMRT physical FMs that can be used to quantitate and rank FMEA severity scoring. Work supported by grants CA10953 and CA81647 (NCI, DHHS).

Flattening filter free beams are either available or are under development on most commercial c‐arm linear accelerators. These beams have many similarities to traditional flattened beams, but have unique characteristics such as much higher dose rates, different profile shapes, and different beam spectra. Consequently, some differences exist in implementation of these beams as compared to traditional flattened beams. Differences exist in the specifics of these beams' interactions with measurement devices and biological systems. These differences, and related recommendations, are in a Therapy Emerging Technology Assessment Work Group document under review by the AAPM. This session is based on that report, and will provide an overview of flattening filter free beams. This will include a technological review, as well as acceptance testing, commissioning, calibration, and periodic quality assurance for these beams. Radiation safety issues (both for patients and personnel) will be addressed, as will radiobiological considerations. Finally, clinical applications and limitations of flattening filter free beams will be highlighted. Learning Objectives: 1. Understand unique modifications to acceptance testing, commissioning, and calibration for FFF beams. 2. Understand unique modifications to periodic QA, as well as patient and personnel safety, appropriate for FFF beams. 3. Understand radiobiological considerations of FFF beams. 4. Understand the clinical applications (strengths and weaknesses) of FFF beams.

Purpose: To analyze the reproducibility of patient specific IMRT QA results across five different clinical devices and one in‐house detector by quantifying variations in the QA results from delivery‐to‐delivery (without re‐setup, "redelivery") and variations when set‐up is repeated (“re‐setup”). Methods: Six patient plans were selected from among four treatment sites. Six methods of IMRT QA were chosen: film and cc04 ion chamber in a body phantom, diode array with all fields AP, as well as with fields delivered rotationally (in a MapPhan phantom), and an in‐house designed solid water phantom with an insert containing 5 ion chambers that can rotate to eight positions. Each of the six plans was delivered three times. The system was then re‐setup and redelivered. This last step was repeated, giving three redelivered plans, and three plans that were delivered under independent set ups. Reproducibility was quantified by the coefficients of variation (CV) of the re‐deliveries and the re‐setup deliveries. Results: Of all detectors investigated, film has the highest CV, with an average CV of 1.56% for re‐deliveries, and 2.46% for re‐setup. For the re‐setup and re‐delivery, the lowest average CV's are the AP diode array with 0.25%, and the ion chamber with 0.13%, respectively. Additionally the standard deviation in dose across the ion chambers calculated by the TPS was compared to the standard deviation of the re‐setup measurements for the in‐house phantom, giving a weak correlation. Conclusion: Different IMRT QA devices exhibit different reproducibility; however all CV's were lower than 5%, indicating good consistency in measurement. Film's dependence on the processor and choice of normalization point are possible reasons why this dosimetry system had a higher CV. This information will help quantify the confidence of IMRT QA performed in the clinic and provide valuable information on the reproducibility of various IMRT QA devices. This work was supported by Public Health Service grants CA010953, CA081647, and CA21661 awarded by the National Cancer Institute, United States Department of Health and Human Services.

Purpose: The Radiological Physics Center (RPC) credentials institutions for clinical trials involving advanced radiotherapy techniques with anthropomorphic phantoms. This study compared phantom results with in‐house IMRT QA results. Methods: We examined RPC head phantom irradiations for which institutions had submitted their own in‐house IMRT QA results for the same plan. The RPC phantom includes 6 TLD‐based point dose measurements, and film for evaluating distance to agreement (DTA) between the target and a neighbouring critical structure. Treatments failed the RPC phantom if any TLD‐measured dose was outside 7% or the DTA in the high gradient region between the target and critical structure was more than 4mm from the institution's TPS‐calculation. Institutions' IMRT QA on the same plan was per their protocol. The institution failed IMRT QA if a) they declared that their results did not meet their internal criteria, or, b) if their point doses agreed outside 3% or planar analysis agreed with less than 90% of pixels passing a 3%/3mm (or less stringent) gamma criteria. 189 plans were examined. Results: 23 plans failed RPC phantom irradiation, while 18 failed in‐house IMRT QA. However, consistency was not observed between the two systems. Plans that failed the RPC phantom irradiation were only 22% likely to be noted as failing in‐house IMRT QA (i.e., sensitivity = 0.22). Plans that passed the RPC phantom tended to pass IMRT QA; the specificity was 92%. No correlation was found between the agreement of in‐house point dose and TPS versus RPC point dose and TPS. Conclusion: Sensitivity, the most important quality of a QA process, was found to be very low overall for in‐house IMRT QA as compared to the RPC phantom. These results challenge the merit of IMRT QA to detect problems in an end‐to‐end treatment delivery. This work was supported by Public Health Service Grants CA10953 and CA081647 awarded by the National Cancer Institute, United States Department of Health and Human Services.

Purpose: To commission a multiple‐source Monte Carlo model of Elekta linear accelerator beams of nominal energies 6MV and 10MV. Methods: A three source, Monte Carlo model of Elekta 6 and 10MV therapeutic x‐ray beams was developed in a two‐step process. Energy spectra of each of three sources, a primary source corresponding to photons created in the target, an extra‐focal source corresponding to photons originating from scattered events in the linac head, and an electron contamination source, were determined. The two photon sources were determined by an optimization process that fit the relative fluence of 0.25 MeV energy bins to the product of Fatigue‐Life and Fermi functions to match calculated percent depth dose (PDD) data with that measured in water for a 10×10cm2 field. Off‐axis effects were modeled by fitting the off‐axis fluence to a piece‐wise linear function through optimization of relative fluence to match calculated dose profiles with measured dose profiles for a 40×40cm2 field. A 3rd degree polynomial was used to describe the off‐axis half‐value layer as a function of off‐axis angle. The model was then commissioned by comparing calculated PDDs and dose profiles for field sizes ranging from 3×3cm2 to 30×30cm2 to those obtained from measurements. Results: Agreement between calculated and measured data was evaluated using 2%/2mm global gamma criterion for field sizes of 3×3, 5×5, 10×10, 15×15, 20×20, and 30×30cm2. Along the central axis of the beam 99.5% and 99.6% of all data passed the criterion for 6 and 10MV models, respectively. Dose profiles at depths of dmax, 5, 10, 20, and 25cm agreed with measured data for 95.4% and 99.2% of data tested for 6 and 10MV models, respectively. Conclusion: A Monte Carlo multiple‐source model for Elekta 6 and 10MV therapeutic x‐ray beams has been developed as a quality assurance tool for clinical trials. This work was supported by Public Health Service grants CA010953, CA081647, and CA21661 awarded by the National Cancer Institute, United States Department of Health and Human Services.

Purpose: To identify non‐small cell lung cancer (NSCLC) CT image feature sets that are reproducible, non‐redundant, and informative across multiple CT machines. Such feature sets will be ideal for modeling and will yield increased generality and robustness to machine variation. Methods: Non‐contrast‐enhanced, test‐retest CT image pairs were obtained from 56 NSCLC patients imaged on three CT machines from two institutions. 328 quantitative image features were extracted from each tumor based on its geometry, intensity histogram, absolute gradient image, co‐occurrence matrix, and run‐length matrix. Feature reproducibility was quantified using the test‐retest data and the concordance correlation coefficient (CCC). The Dice similarity coefficient (DSC) and the Jaccard index (JI) were used to quantify reproducible feature set agreement between machines. Multi‐machine reproducible feature sets were created by taking the intersection of individual machine reproducible feature sets. Redundant features were removed through hierarchical clustering based on the average correlation between features across multiple machines. Results: Across machines, the average percentage of features that were reproducible (CCC>0.90) was highest for 4D‐CT average images (92.5%), intermediate for 4D‐CT end‐exhale images (73.0%), and lowest for 3D‐CT breath‐hold images (61.0%). For 4D‐CT, end‐exhale image features were relatively machine‐sensitive(DSC = 0.71, JI = 0.55), and average image features were relatively machine‐insensitive (DSC = 0.90, JI = 0.87). Multi‐machine reproducible and non‐redundant feature sets were identified for 4D‐CT average images, 4D‐CT end‐exhale images, and 3D‐CT breath‐hold images. Hierarchical clustering confirmed feature set informativeness: 55 out of 56 test‐retest images were correctly paired using only their reproducible, non‐redundant feature set values. Conclusion: Image feature reproducibility was machine‐sensitive. High quality machine‐robust feature sets were identified for each image type. Compared to end‐exhale 4D‐CT and breath‐hold 3D‐CT, average 4D‐CT derived image features showed superior multi‐machine reproducibility and are the best candidates for clinical correlation.

Purpose: We set out to revise/improve the Radiation Detection Instrumentation and Data Analysis course at our institution due to negative student evaluations of this course. Initially, we began with changes to the syllabus, adding important topics that were not being covered and expanding the teaching faculty to include board‐certified physicists with practical clinical experience. Although this led to improved student course evaluations compared to the previous years, we were still not satisfied that enough material was being covered in the course. Because the time in the semester was limited, we considered alternative teaching methodologies in an effort to broaden the scope of the course. Methods: We implemented project‐based learning because it allowed us to keep instructor‐led lectures, which provide the students with fundamental knowledge that we believe essential for a core course. We also expanded the breadth of the topics covered in the course through assignment of a class project. The class project broadly aims to provide opportunities for each group of students to investigate meaningful real‐world topics related to radiation detection in medical physics. For this assignment, the students assume the role of director of the medical physics radiation therapy department at a comprehensive cancer treatment center and are asked to develop a capital equipment list of detectors and associated equipment to perform all necessary QA for this department. They are expected to submit a detailed report and give a presentation to the class. Results: The class project was first assigned in 2010 and has been required every year since. The project has increased the breadth of the course and student course evaluations have improved (82% positive compared to 36% positive). Conclusion: Overall, our students are learning more material in the revamped course and are exposed to one of the many facets of a career in clinical medical physics.

Purpose: To analyze the sensitivity of patient‐specific IMRT QA devices to detecting plan failures when compared to multiple ion chamber measurements. Methods: Four methods of IMRT QA were chosen: film (Kodak EDR2) and ion chamber (Wellhofer cc04) in an I'mRT body phantom, MapCheck2 (Sun Nuclear) with fields delivered AP, and MapCheck2 in the MapPhan phantom with rotational fields. Additionally, an in‐house designed multiple ion chamber phantom with an insert containing 5 ion chambers (Exradin A1SL 0.057cc) that can rotate to eight positions was used as a gold standard. Twenty plans previously failing film and ion chamber IMRT QA at our institution and five passing plans were delivered to all devices. The plans' performance on the in‐house phantom sorted them into true fails and passes. The sensitivity to detect true failures was assessed across devices with acceptance criteria of 3%/3mm and 90% of pixels passing for planar, and 3% dose difference for point measurements. Additionally, specificity was analyzed. Results: The sensitivities among all devices were less than 0.70 (poor), indicating a deficiency in identifying plans that failed versus multiple ion chamber measurements. The specificities were generally high. The AP‐delivered MapCheck has the lowest sensitivity (0.27), while the rotationally‐delivered Mapcheck MapPhan had the lowest specificity (0.70) due the directional dependence of diodes. Film was the most sensitive (0.60) dosimeter under investigation, while the single ion chamber had the highest specificity (1.00). Conclusion: The different sampling methods of each device lead to a difference in the ability to detect true failures as determined by multiple ion chamber readings. The overall low sensitivities among all devices indicates that failing plans are passing, leading to a potential lack of error detection in patients' IMRT plans. This work was supported by Public Health Service grants CA010953, CA081647, and CA21661 awarded by the National Cancer Institute, United States Department of Health and Human Services.

Purpose: Use optically stimulated luminescent dosimeters (OSLD) whose visible active luminescent area were masked as a remote audit tool to measure small field photon doses down to a 7.5mm diameter field size. Methods: Different aperture mask sizes (1, 2, and 3mm) were made for the OSLD nanodots and tested to determine which had the best reproducibility. These masks defined the active area of the OSLDs. The OSLDs were characterized by determining correction factors for — linearity, fading, depletion, element, and energy, in order to calculate dose. OSLD readings were performed with an OSLD reader with a known LED light distribution. Dose rate versus field size for OSLDs were generated and compared with other dosimeters (film, diodes, ion chamber, and Monte Carlo). All measurements were performed using a 6MV photon beam. Results: The study determined the 2mm diameter aperture to be the smallest aperture with the best reproducible readings between each individual OSLD with coefficient of variation ranging between 1–1.5%. Linearity tests for the masked OSLDs showed correspondence with unmasked commissioning data to within 3% for both the 2mm and 3mm diameter aperture, for doses ranging from 50cGy to 200cGy. The 1mm masked OSLD will not be further investigated as it produced non‐reproducible results and its linearity correction was not in agreement with the commissioning unmasked correction. Conclusion: The 2mm masked OSLDs can potentially be used as a remote audit dosimetry tool to measure small photon field sizes doses. This work was supported by Public Health Service grants CA010953, CA081647, and CA21661 awarded by the National Cancer Institute, United States Department of Health and Human Services.

Purpose: This study examined MLC positional accuracy via MLC logs from multiple institutions and multiple delivery techniques to evaluate typical positional accuracy and treatment and mechanical parameters that affect accuracy. Typical accuracy achieved was compared against TG‐142 recommendations for MLC performance; more appropriate recommendations are suggested. Methods: Over 85,000 Varian MLC treatment logs were collected from six institutions and analyzed with Fraction CHECK. Data were binned according to institution and treatment type to determine root mean square (RMS) and 95th percentile error values, and then to look for correlations between those errors and with mechanical and treatment parameters. Results: Results of treatment logs from six institutions found that leaf RMS error and 95th percentile leaf error were fairly consistent between institutions, but varied by treatment type. The step and shoot technique had very small errors: the mean RMS leaf error was 0.008 mm. For dynamic treatments the mean RMS leaf error was 0.32 mm, while VMAT showed the largest mean RMS leaf error at 0.46 mm. For the dynamic and VMAT techniques, the mean and maximum leaf speeds were significantly linked to the leaf RMS error. For dynamic delivery, the mean leaf error was correlated with RMS error, whereas for VMAT the average gantry speed was correlated. For all treatments, the RMS error and the 95th percentile leaf error were correlated. Conclusion: Restricting the maximum leaf speed can help improve MLC performance for dynamic and VMAT deliveries. Furthermore, the tolerances of leaf RMS and error counts for all treatment types should be tightened from the TG‐142 values to make them more appropriate for clinical performance. Values of 1 mm for the 95th percentile of leaf RMS error and 1.5 mm for the 95th percentile error are suggested as action levels for all treatment types.

Purpose: To investigate the dosimetric impact of implementing two improvements to the convolution/superposition (C/S) method that are generally ignored in traditional implementations of the algorithm. First, the impact of taking into account spectral changes of the incident photon beam in the energy deposition kernel (EDK) calculation (kernel hardening method), rather than using a single polyenergetic kernel that reflects the spectrum at a single location, was investigated. Second, the impact of implementing material‐specific kernels, rather than performing density scaling of water kernels, was investigated. Methods: To investigate the impact of kernel hardening, depth dose curves were calculated for a clinical 6MV photon beam incident on a water phantom using two different implementations of the collapsed cone C/S method (Mobius Medical Systems, Houston, TX), one using a single polyenergetic kernel and one that fully takes into account spectral changes in the kernel calculation. To investigate the impact of material‐specific kernels, depth dose curves were calculated for a simplified titanium implant geometry (4×4×4 cm3 titanium cavity inside a water phantom) using both a traditional C/S implementation that performs density scaling of water kernels and a novel implementation using titanium kernels (generated using the EGSnrc user code EDKnrc). Results: Implementation of kernel hardening increased the PDD value at 25 cm depth by 2.1% to 5.8% depending on the field size. Implementation of titanium kernels gave a 1.5% higher dose upstream of the metal cavity (i.e., higher backscatter dose) and a 5.9% lower dose downstream of the cavity. Conclusion: Implementation of kernel hardening affected the shape of the C/S‐calculated depth dose curves and thus has the potential to affect beam modeling parameters obtained in the commissioning process. For metal implants, the C/S algorithms generally underestimate the dose upstream and overestimate the dose downstream of the implant; implementation of material‐specific kernels mitigated both of these errors. This work was supported by Public Health Service grants CA010953, CA081647, and CA21661 awarded by the National Cancer Institute, United States Department of Health and Human Services.

Purpose: Investigate the error in IMRT dose delivery from changes in beam quality, symmetry, and MLC positional errors. Methods: A complex (3533 MU, 216 segments, 0.181 modulation complexity score) H&N IMRT plan was delivered to the RPC's IMRT H&N phantom on a Varian Clinac 2100CD. The plan was delivered as a baseline and then again after adjusting the in‐plane and cross‐plane symmetry by 5% and 3%, respectively. The beam was adjusted back to baseline performance and the beam quality was hardened by nearly 2% (%dd) by altering the bending magnet current. The beam was then readjusted to baseline and the plan was delivered with MLC offsets of +1 mm and +3 mm. Radiochromic film and TLD within the phantom were used to analyze the 2D dose distributions and absolute doses. Results: Relative to the baseline delivery, differences in absolute dose delivery of up to 0.6% and 3.2% were observed with changes in in‐plane and cross‐plane symmetry, respectively. The beam symmetry changes gamma analyses (±7%/4mm) resulted in a 0.4–11% change in the percent of pixels passing. The small change in beam quality resulted in a 6% average absolute dose and 24% pixel pass rate difference. MLC offsets of 1 mm and 3mm resulted in average absolute dose differences of 12% and 32%, respectively, and a gamma analyses 21% and 33% change in pixel pass rate for each offset, respectively. Conclusions: Accurate IMRT dose delivery is a long, complicated chain of events, each of which might contribute to dose delivery errors. Performing a risk‐based FMEA analysis (presented by TG‐100) to establish a QA program for IMRT requires knowledge of the error magnitude in dose delivery from key failure modes. Beam quality and MLC offsets resulted in the largest dose delivery errors. Additional failure modes are being investigated. Work supported by grants CA10953 and CA81647 (NCI, DHHS).

Purpose: To obtain and assess the accuracy of 3D dose distributions and LET maps from a passively‐scattered proton beam through the use of LET‐dependent and LET‐independent polymergel formulations. Methods: Dose measurements were performed with BANG3‐Pro2 (MGS Research, Inc, Madison, CT)polymergel dosimeters. The dosimeters were mixed from kits and were poured into 14.5 cm high by 15 cm diameter acrylic cylinders. For LET measurements, a new BANG3‐Pro variant intended for use as an ‘LET‐ meter’ was employed. Initial irradiations were performed using a 200 MeV passively scattered proton beam with a 4 cm spread out Bragg peak (SOBP), delivering a physical dose of 3 Gy to the center of the SOBP. Dosimeters were read out using the OCTOPUS‐IQ (MGS Research, Inc, Madison, CT) optical CT scanner. The optical densities measured in the gel were compared against ion chamber data to assess the accuracy of the dosimeter. Results: Initial analysis indicates 80% of points measured along the central axis agreed with ion chamber data at the ±5%/±3mm level in the sensitive region of the gel. An artifact in the reconstructed image produced inaccurate readings beyond the depth of 50% dose on the distal edge of the SOBP, resulting in the majority of the agreement failures. Conclusions: BANG3‐Pro2 polymergel dosimeters demonstrate promise as a 3D dosimeter for use in proton therapy. The dosimetric data obtained to date will be used as the baseline measurement against which the LET‐sensitive BANG3‐Pro formulation will be compared for the measurement of proton LET.

Purpose: To investigate the effects of image noise and resolution on gamma analysis results for clinical IMRTquality assurance (QA) treatment plans. Methods: For each clinical IMRT plan, a hybrid QA plan was created in Pinnacle3 using a solid water phantom and delivered with a Varian Clinac 21EX. A transverse dose plane was measured using Kodak EDR2 radiographic film. Gamma analyses were performed using OmniPro‐I'mRT software (IBA Dosimetry, Germany), comparing the calculated dose distribution (reference) versus the measured dose distribution (evaluated). Each IMRT QA film was digitized using three different image resolutions (71, 142, and 285 dpi). In order to study the potential for noise (e.g., from the film, developer, or digitizer) to mask an error, a 4 mm positional shift and various noise levels (normally distributed with 1, 2, and 3% standard deviation) were introduced to the calculated dose distributions, and gamma analyses were performed using these noisy distributions as the measured dose distributions. Results: Our analysis showed that the percentage of passing pixels increased with both increasing image resolution and image noise. Doubling the film image resolution increased the passing rate by an average of 0.1% for ±5%/3mm acceptance criteria, 0.8% for 3%/3mm, and 2.6% for 2%/2mm for the six clinical IMRT QA plans in this study. Increasing the noise from 0% to 3% increased the passing rate of the shifted images by an average of 1.5% for the 5%/3mm acceptance criteria, 4.6% for 3%/3mm, and 8.9% for 2%/2mm. Conclusions:Image noise and high scanning resolution artificially increased the percentage of passing pixels in gamma analysis. This effect was more pronounced for tighter γ criteria (i.e. 2%/2mm). In designing IMRT QA protocols, it is important to be aware of the fact that gamma analysis is sensitive to these two parameters.

Purpose: Ion recombination is approximated in TG‐51 by Pion, which is calculated by a 2‐voltage measurement. This approximation is sufficiently accurate for the calibration of beams only when Pion is less than 1.05. High‐dose per pulse beams, such as flattening filter free beams, may increase Pion to an unacceptable level for the 2‐voltage measurement approach to be adequate. Methods: Pion was measured for flattened beams of 6, 10, 15, and 18 MV, as well as flattening filter free beams of 6 and 10 MV. Pion was also measured for electron beams of 6–18 MeV at standard as well as high dose rate (1000 MU/min). The measurements were made at a range of depths and with a variety of clinical ion chambers, including PTW, NEL, and Exradin chambers. 1/V versus 1/I curves were generated for the scenarios generating the maximum measured Pion values in order to confirm the accuracy of the 2‐voltage technique. Results: Consistent with the increased dose per pulse, Pion was higher for FFF beams than for FF beams. However, for all beams, measurement locations, and chambers examined, Pion never exceeded 1.018. Compared to the 1/V versus 1/I results, the 2 voltage technique was always accurate within 0.3%. Conclusions: Recombination can be adequately accounted for in high dose rate flattening‐filter free beams using the standard 2‐voltage technique to determine Pion.

Purpose: To analyze the findings of the Radiological Physics Center's (RPC) QA audits of institutions participating in NCI sponsored clinical trials. Methods: The RPC has developed an extensive Quality Assurance (QA) program over the past 44 years. This program includes on‐site dosimetry reviews where measurements on therapy machines are made, records are reviewed and personnel are interviewed. The program's remote audit tools include mailed dosimeters (OSLD/TLD) to verify output calibration, comparison of dosimetry data with RPC ‘standard’ data, evaluation of benchmark and patient calculations to verify the treatment planning algorithms, review of institution's QA procedures and records, and use of anthropomorphic phantoms to verify tumordose delivery. The RPC endeavors to assist institutions in finding the origins of any detected discrepancies, and to resolve them. Results: Ninety percent of institutions receiving dosimetry recommendations has remained level for the past 5 years. The most frequent recommendations were for not performing TG‐40 QA tests, wedge factors, small field size output factors and off‐axis factors. Since TG‐51 was published, the number of beam calibrations audited during visits with ion chambers, that met the RPC's ±3% criterion, decreased initially but has risen to pre‐TG‐51 levels. The OSLD/TLD program shows that only ∼3% of the beams are outside our ±5% criteria, but these discrepancies are distributed over 12–20% of the institutions. The percent of institutions with i,3l beam outside the RPC's criteria is approximately the same whether OSLD/TLD or ion chambers were used. The first time passing rate for the anthropomorphic phantoms is increasing with time. The prostate phantom has the highest pass rate while the spine phantom has the lowest. Conclusions: Numerous dosimetry errors continue to be discovered by the RPC's QA program and the RPC continues to play an important role in helping institutions resolve these errors. This work was supported by PHS grants CA10953 and CA081647 awarded by NCI.

Purpose: To evaluate normal tissuedose reduction in step‐and‐shoot intensity‐modulated radiation therapy(IMRT) on the Varian 2100 platform by tracking the multileaf collimator(MLC) apertures with the accelerator jaws. Methods: Clinical radiation treatment plans for 10 thoracic, 3 pediatric and 3 head and neck patients were converted to plans with the jaws tracking each segment's MLC apertures and compared to the original plans in a commercial radiation treatment planning system (TPS). Each segment was then renormalized to account for the change in collimator scatter to obtain target coverage within 1 % of that in the original plan. Reduction in normal tissuedose was evaluated in the new plan by using the parameters V5, V10, and V20 in the cumulative dose‐volume histogram for the following structures: total lung minus GTV (gross target volume), heart, esophagus, spinal cord, liver, parotids, and brainstem. In order to validate the accuracy of our beam model, MLCtransmission measurements were made and compared to those predicted by the TPS. Results: The greatest change between the original plan and new plan occurred at lower dose levels. The reduction in V20 was never more than 6.3% and was typically less than 1% for all patients. The reduction in V5 was 16.7% maximum and was typically less than 3% for all patients. The variation in normal tissuedose reduction was not predictable, and we found no clear parameters that indicated which patients would benefit most from jaw tracking. Our TPS model of MLC transmission agreed with measurements with absolute transmission differences of less than 0.1 % and thus uncertainties in the model did not contribute significantly to the uncertainty in the dose determination. Conclusion: The amount of dose reduction achieved by collimating the jaws around each MLC aperture in step and shoot IMRT is probably not clinically significant.

Purpose: Optically stimulated luminescent dosimeters (OSLD) are becoming more popular for dose measurement in a clinical setting. The OSLD response is dependent on photon energy and that energy response is typically determined only for the photon spectrum at the reference dose calibration point (dmax on CAX for a 10cm×10cm field). Previous work has shown that variations exist in photon energy spectra as a result of measurement conditions and treatment parameters; however the effects of these energy variations on OSLD response have not been characterized. Method and Materials:Theoretical energy correction factors were calculated for a range of clinical conditions, including 6MV static and modulated (IMRT) fields, in‐field and out‐of‐field measurement positions, and in the presence of heterogeneous materials. These factors were calculated using previously determined photon energy spectra and Burlin cavity theory. Measured energy correction factors using OSLD nanoDots™ from Landauer, Inc. have also been determined under matching conditions, and the measured and calculated responses were compared. Results: When OSLD are used to measure dose for in‐field locations, the dosimeter response was as much as 5% different from the reference location due to perturbations in the spectra. The presence of heterogeneities at in‐field measurement locations did not significantly impact the OSLD response; however a substantial energy response occurred for the soft spectra that exist outside of the treatment field. At these measurement locations, OSLD may over‐respond by 20% or more, relative to the response at the dose calibration location. Conclusion: OSLD exhibit a non‐trivial energy response due to the increased effective atomic number. Variations in the photon energy spectra may impact the response of OSLD by as much as 20% and additional non‐reference energy correction factors may be necessary when measuring dose away from the dose calibration position. Work supported by PHS grant CA 10953 awarded by NCI, DHHS

Purpose: To evaluate the dosimetric impact of the Varian Exact Couch top and rails on dose and relative volume coverage to target and critical structures for 6 MV IMRT and Arc Therapy. Methods: Five prostate patients were planned with both 6 MV 8‐field IMRT and 2‐field RapidArc per clinical practice at MDACC for target and normal tissue constraints using the Eclipse treatment planning system. The dose distributions were then re‐calculated with inclusion of the couch top and rails in varying configurations. DVH analysis of the target and critical structures for dose and relative volume coverage loss were evaluated and compared to MDACC clinical plan requirements (98% and 95% coverage of prostate and PTV, respectively, at prescription dose). To isolate the effects of the couch top and rails individually, the dose distribution was re‐calculated for the treatment plans with the couch top only. Calculated dose perturbations were verified through measurements in an IMRT QA phantom. Results: The average dose loss to the prostate and relative volume coverage (at prescription dose) were: 3.2 Gy/35% and 1.5 Gy/84% for IMRT with rails out and in, respectively, and 2.4 Gy/18% and 2.2 Gy/17% for RapidArc with rails out and in, respectively. The couch top accounted for 1.5 Gy/84% for IMRT and 1.5 Gy/40% for RapidArc, the remainder was from the rails. Conclusions: Dose and volume coverage loss for IMRT plans is primarily due to the rails while the couch top is the primary cause for RapidArc. Both the couch top and rails contribute to dose and coverage loss to a degree that, if included, would cause the plan to fail clinical approval. Therefore, the couch top and rails should be accounted for in treatment planning.

Purpose: To develop and test the sensitivity of a single exposure film calibration procedure using a dynamic wedge. Method and Materials:A 60o dynamic wedge was used to expose EDR2 film and generate optical density (OD) to dose calibration curves in DoseLab Pro. Doses between 50 and 500 cGy were examined. Dosimetric data was generated from a Profiler2 diode array. This calibration curve was compared to one generated by the standard approach of irradiating 8 separate sheets of film to known doses. The sensitivity of the wedged‐film calibration procedure was also tested through the following 4 investigations. First, film measurements were repeated 10 times to determine the variability of the film response. Second, profiler measurements were repeated 10 times to determine the variability of the profiler response. Third, positional offsets of the profiler were investigated (up to 4mm off‐center — the size of the diode spacing). Fourth, different amounts of solid water build‐up were used (between 1 and 10cm) above the profiler and corresponding film. Results: The calibration curve generated with the wedge exposure compared very well to that generated with the standard 8‐film approach, being less than 2.5% locally different (0.8% relative to maximum) at doses above 50 cGy. Irradiation of multiple films introduced a maximum deviation of 1.2%, while multiple profiler measurements introduced a maximum deviation of 0.5%. Shifts in the position of the profiler introduced a maximum error of 3.0%. These deviations represent good robustness of the calibration procedure. Consistent calibration curves were generated regardless of the extent of build‐up material used. Each film exposure and software analysis was completed in 1 minute and 5 seconds, respectively. Conclusions: This work has developed a single exposure film calibration procedure. This offers physicists a rapid and robust film calibration method.

Purpose: Investigate whether increased intensity modulated radiation therapy(IMRT)treatment plan complexity or mismatched accelerator dosimetry data are responsible for or contribute to failures with the Radiological Physics Center's IMRT H&N QA phantom. Methods: Eight H&N IMRT plans with a range of total MU (1460–3466), number of segments (54–225), and modulation complexity scores (MCS) (0.181–0.609) were created in Pinnacle v.8 and delivered to the RPC's H&N phantom on a single Varian Clinac. One of the IMRT plans (1851 MU, 88 segments, and MCS=0.469) was equivalent to the median H&N plan from ∼1000 RPC H&N phantom irradiations. This average IMRT plan was also delivered on four matched Varian Clinac machines and the dose distribution calculated using a different 6MV beam model. Radiochromic film and TLD within the phantom were used to analyze the dose profiles and absolute doses. Results: Increasing the treatment plan complexity by varying the MU, number of segments, or the MCS resulted in no clear trend toward an increase in dosimetric error using either ±7%/4mm or ±5%/3mm gamma index criteria. Varying the delivery machines as well as the beam model (use of a Clinac 6EX 6MV beam model vs. Clinac 21EX 6MV model), also did not show any clear trend towards an increased dosimetric error using the criteria indicated above. Conclusions: Accurate IMRTdose delivery requires a complicated chain of events involving numerous steps, each of which might contribute to dose delivery errors. IMRT modulation complexity and mismatched dosimetry data have been postulated as contributing to IMRTdose delivery errors but our results indicate otherwise. Other components of the IMRT delivery chain such as MLC performance and modeling uncertainty in setup, machine mechanical integrity, etc. should be investigated to determine whether they contribute significantly to IMRTdose delivery errors. Work supported by PHS CA010953 and CA081647, awarded by NCI, DHHS.

Purpose: A factor that greatly limits the use of Monte Carlo methods for patient specific dose calculations in research is the substantial amount of effort required to define the Monte Carlo geometry of the actual treatment and patient setup. The purpose of this study was to develop and validate computational infrastructure to automatically convert radiation field parameters and computed tomography(CT) data from Digital Imaging and Communications (DICOM)format to Monte Carlo input format, and automate a Monte Carlo based dose calculation system for external beam photonradiation therapy. Methods: Computational infrastructure was developed using DCMTK, the DICOM tool kit. The dose calculation system (ADCS) was automated using a shell script. For validation of the ADCS, infield doses calculated by the ADCS were compared with those calculated by the treatment planningsoftware using an eight‐field, 6 MV, step‐and‐shoot intensity modulated radiation therapy plan. Doses were calculated in a water phantom, and an anthropomorphic phantom based on CT data. Results: For each field, more than 95% of the dose voxels passed 3% and 3‐mm criteria of gamma‐index analysis of 3‐D dose distributions in the water phantom. The central axis depth dose curve agreed with in 15% (1stardard deviation) to the TPS calculations in the anthropomorphic phantom for a single representative field of the RT plan. Conclusions: The ADCS can accurately extract patient‐specific radiation treatment parameters and automatically incorporate them into the Monte Carlo format and create final dose distributions compatible with commercial treatment planning systems. This automated calculation infrastructure reduces the time required to define the Monte Carlo geometry and potential for human error. This system is capable of automating and calculating doses for conventional radiation therapy and 18MV. It is also capable of calculating out‐of‐field doses (region of interest in patient anatomy can be selected prior to running ADCS).

Purpose: Optically stimulated luminescencedetectors (OSLDs) are quickly gaining popularity as passive dosimeters, with applications in medicine for linac output calibration verification, brachytherapy source verification, treatment plan quality assurance, and clinical dose measurements. With such wide applications, these dosimeters must be characterized for numerous factors affecting their response. The purpose of this study was to examine the angular dependence of the nanoDotOSLdosimeter, which is part of the InLight series from Landauer, Inc. Methods: Relative dosimeter response data were taken at several angles in 6 and 18 MV photon beams, as well as in clinical proton beams. These measurements were done within a phantom at a depth beyond the build‐up region. Measurements were conducted over several sessions to provide confidence in the data. To examine the reasons for the observed angular dependence Monte Carlo simulations were performed with MCNPX. Results: In photon beams, the nanoDot had an angular dependence of 4% at 6 MV and 3% at 18 MV respectively, at angles where the dosimeter was parallel to the beam. Monte Carlo simulations at 6 MV showed similar results to the experimental values. In proton beams, no angular dependence was found. Conclusions: A significant angular response of this OSLD was observed in photon beams. This factor must be accounted for when evaluating doses from photon beams impinging from non‐normal directions.

Purpose: To develop a modified Winston‐Lutz test for determining Image Guided Radiotherapy(IGRT) setup accuracy relative to the radiation isocenter. Methods: A Winston‐Lutz cube (Modus Medical Devices) was placed directly on the treatment table and offset from isocenter by 5.0 mm in the longitudinal, lateral, and vertical dimensions. A high resolution Cone‐Beam CT(CBCT) was acquired and aligned to the reference CTimage. The table was shifted and Anterior‐Posterior (AP) and Lateral (LAT) Electronic Portal Imaging Device(EPID)images were acquired, with radiation to cube isocenter agreement determined for each image. This procedure was repeated ten times to determine reproducibility, and then was repeated daily for 51 days. In addition, this test was repeated daily with a random cube shift. Results: The reproducibility test yielded a mean 3D vector isocenter displacement of 0.7 mm ± 0.23 and 0.26 for the AP and LAT images, respectively. Maximum deviation was 1.09 mm for the AP images and 1.1 mm for the LAT images. The daily 5.0 mm offset test yielded a mean displacement of 0.92 mm ± 0.38 with a maximum of 1.7 mm for the AP images and 0.95 mm ± 0.47 with a maximum of 2.0 mm for the LAT images. The random offset test yielded a mean displacement of 0.68 mm ± 0.31 with a maximum of 1.54 mm for the AP images and 0.89 mm ± 0.38 with a maximum of 2.0 mm for the LAT images. Conclusions: A modified Winston‐Lutz test was developed that helps to quantify the maximum error to be expected for IGRT based setup patients. The test took into account inaccuracies due to IGRT isocenter definition, MV isocenter variation, IGRT software alignment, IGRT couch motions, TPS centroid isocenter definition, CT slice width effects, ROI contouring accuracy, and MLC positioning.

Purpose: First, to determine an average and maximum displacement of the shoulder relative to isocenter over the course of treatment. Second, to establish the dosimetric effect of shoulder displacements relative to correct isocenter alignment on the dose delivered to the target and the surrounding structures for head and neck cancer patients. Methods: The frequency of shoulder shifts of various magnitudes relative to isocenter was assessed for 4 patients using image registrationsoftware. The location of the center of the right and left humeral head relative to isocenter (usually C2) was found daily from CT on rails scans, and was compared to the location of the humeral heads relative to isocenter on the initial simulation CT. Three Baseline head and neck IMRT and SmartArc plans were generated in Pinnacle based on simulation CTs. The CT datasets (external contour and boney structures) were then modified to represent shifts of the shoulder (relative to isocenter) between 3 mm and 15 mm in the SI, AP, and LR directions. The initial plans were recalculated on the image sets with shifted shoulders. Results: On average, shoulder motion was 2‐5 mm in each direction, although displacements of over 1 cm in the inferior and posterior directions occurred. Shoulder shifts induced perturbations in the dose distribution, although generally only for large shifts. Most substantially, superior shifts resulted in coverage loss of up to 152 cc for targets in the lower neck. Inferior shifts elevated the dose to the brachial plexus by 0.6–4.1 Gy. SmartArc plans showed similar loss of target coverage as IMRT plans. Conclusions: The position of the shoulder can have an impact on target coverage and critical structure dose. Shoulder position may need to be considered for setup of head and neck patients depending on target location.

Purpose: Optically stimulated luminescent detectors (OSLDs) have a number of advantages in radiationdosimetry making them excellent dosimeters for quality assurance and patient dose verification. Although the dosimeters have been investigated in several modalities, relatively little work has been done in examining the dosimeters for use in clinical proton beams. This study examined a number of characteristics needed before the dosimeters should be used clinically in proton beams. Methods: OSLdosimeters from Landauer, Inc. specifically the nanoDot, were investigated. These dosimeters were placed in a special phantom with recesses to fit the dosimeters with minimal air gap. Beams with nominal energies of 160, 200, and 250 MeV were used in the passively‐scattered proton beam at the MD Anderson Cancer Center Proton Therapy Center. SOBP measurements were always done at the center of the SOPB. Results: The dosimeters showed supralinearity with doses above 200 cGy, with 5% supralinearity at 1000 cGy. Field size dependence was measured from sizes 2×2cm2 to 18×18cm2, with no noticeable dependence. Residual signal was measured as a function of cumulative protondose, showing only a small increase up to 800 cGy. Readout signal depletion of the dosimeters after consecutive readings was done for various protondose levels and compared to photons. While the depletion differed slightly for protons, no dose dependence was found. Energy dependence of the dosimeters was examined in the center of various SOBP widths and energies, showing general agreement with ion chamber data with one exception. This exception is being investigated. Conclusion: The nanoDotOSLdosimetercharacteristics were studied in clinical proton beams and found to have dose supralinearity above 200 cGy, and no dependence on field size, or significantly different readout signal depletion with dose. Residual readout signal with cumulative dose did not increase largely with dose.

Purpose: Current dose‐response studies use symptom endpoints (e.g. grade 2 esophagitis) as substitutes for data about actual tissue damage. CTimaging could provide objective spatial data on radiation‐induced damage to the esophagus in lungcancer patients. Methods: Deformable image registration techniques were used to register weekly CTimages taken during radiotherapytreatment with the original planning CTimage. The esophagus contours were automatically mapped. The impact of day‐to‐day variations in the degree of collapse of the esophagus was reduced by using thresholding to remove air from the esophagus. The cross‐sectional area of the esophagus was then calculated for each CT slice. The results from surrounding slices were averaged to reduce apparent abrupt steps in cross‐ sectional area between adjacent CT slices caused by varying undulations/folds in the esophagus. Finally, the relative expansion of the esophagus was calculated as the ratio of the cross‐sectional area of the esophagus in the weekly CT (minus air) to that on the corresponding CT slice in the planning image. This technique was applied to weekly CTimages of 5 lungcancer patients (35 CTs), with acute esophageal toxicity grade 0 to 3. For these patients we examined (1) The correlation between the relative expansion of the esophagus and the clinical toxicity grade, and (2) the correlation between the spatial dose distribution and the spatial variation in esophageal expansion. Results: The average maximum esophageal expansions for toxicity grades 0, 2, and 3 were 1.2, 1.7 and 1.9, respectively. The difference between grade 2 and 0 was statistically significant (p=0.008). The location and degree of variation of the changes in esophagus cross‐section were found to be related to the high dose given at the same location. Conclusion: Radiation‐induced injury to the esophagus can be detected in CTimages. This has potential for use in dose‐response studies.

Purpose: To test the sensitivity of the Eclipse anisotropic analytical algorithm (AAA) to expanded commissioning data that includes small‐field data and data corrected for detector size. Method and Materials: A 0.01 cm3ionization chamber was used to measure an extended commissioning data set for a 6 MV photon beam including open field profiles (OFP), percent depth dose (PDD), and output factors for field sizes ≥ 1×1 cm2. OFP were processed using mean‐value symmetry and adjusted for detector size effects. OFP were smoothed outside the penumbra region for noise reduction. PDD data were shifted to account for the effective point of measurement and then smoothed. Four unique beam models were commissioned to evaluate the impact of correcting the measured data for detector size as well as small‐field data inclusion. We compared the resulting AAA calculated models with the model based on Golden Beam Data (GBD) provided by Varian Medical Systems. Models were compared using a 1 mm calculation grid. Results: The various models proved to be insensitive to adjustments for detector size effects and also to the inclusion of small‐field data. All models, regardless of the penumbra of input profiles, exhibited nearly identical penumbra in the final calculated model. The algorithm narrowed the penumbra for the model commissioned with GBD i.e. with broad penumbra. Conversely, the algorithm broadened the penumbra for the model commissioned with measured data that were smoothed and corrected for volume averaging effects i.e. data with narrow penumbra. Conclusion: Results suggest that the TPS utilizes a generic model for the penumbra region in calculated models, thus ignoring measured data. The negligible impact of inclusion of small‐field data on the TPS's calculated model may be due to the large relative size of the penumbra region for those fields.

Purpose: Validation of dose calculation accuracy is essential for new radiotherapy techniques, such as volumetric modulated arc therapy (VMAT). The purpose of this project was to compare a VMAT system, Varian RapidArc, to the current standard of care, IMRT, using Radiological Physics Center (RPC) phantoms in terms of both treatment plan quality and dosimetricdelivery accuracy. Method and Materials:Treatment plans and dosedelivery accuracy were evaluated following procedures defined by the RPC for credentialing institutions for Radiation TherapyOncology Group (RTOG) clinical trials. Clinically relevant treatment plans were created for RPC prostate and head and neck phantoms from typical prescription and dose constraints for 6 MV IMRT and RapidArc. The treatment plans for head and neck radiotherapy were compared to determine if they were clinically comparable using several dosimetric criteria, including ability to meet dose objectives set by the RPC and conformity and homogeneity indices. The planned treatments were delivered to the phantoms and absolute doses and relative dose distributions were measured with thermoluminescent dosimeters and radiochromic film, respectively. The measured and calculated doses of each treatment plan were compared to determine if they were clinically acceptable based upon dose differences and distance‐to‐agreement. Results: All treatment plans were able to meet the dosimetric objectives set by the RPC for treatment volume dose coverage and critical structure dose constraints for both the prostate and head and neck phantoms. The IMRT and RapidArc plans for the head and neck phantom had similar conformity and homogeneity. The RapidArc plan for the prostate phantom met the RPC delivery accuracy requirements. Preliminary results for the head and neck phantom indicated that the IMRT and RapidArc plans were delivered with similar accuracy. Conclusion: Varian RapidArc is comparable to IMRT in treatment plan quality and the dose calculation accuracy is clinically acceptable.

Dose outside of the target volume is an unavoidable consequence of radiation therapy. This undesirable radiation is clearly a concern for pregnant patients as the fetus is highly radiosensitive. Similarly, out‐of‐field radiation is a concern for patients with implanted cardiac pacemakers or defibrillators as electronic failures may be induced by relatively low doses of radiation. Out‐of‐field dose has also received attention due to its potential for inducing late effects such as second malignant neoplasms,heart disease, or stroke. In recent years patient survival has increased due to more efficacious treatments resulting in more long‐term survivors. These long‐term survivors have an increased risk of developing a radiation induced late effect. Compounding this, modern radiotherapy techniques, such as intensity‐modulated radiation therapy(IMRT) may increase the out‐of‐field dose, increasing the risk per person of developing a late effect. The above concerns surrounding out‐of‐field dose are confounded by the number and complexity of modern radiotherapy delivery techniques, such as IMRT, Tomotherapy, Cyberknife, and Gamma Knife treatments. Out‐of‐field dose is not accurately reported by commercial treatment planning systems and other methods for assessing this dose are necessary. Therefore, measurements or calculations are often performed to evaluate out‐of‐field doses. As there are a variety of measurement and calculational options available, it is worthwhile exploring the strengths and limitations of each, as well as particular cautions that may be required. Monte Carlo calculations may be much more accurate than treatment planning system calculations, but require substantial infrastructure and modeling. Measurements may be a reasonable approach, but it is nevertheless possible to have substantial error in the measurement, particularly with TLD. This presentation will examine the out‐of‐field dose associated with the variety of different treatment modalities that are currently available. The sources and relative importance of head leakage, collimator scatter, and patient scatter will be compared, and parameters that impact the out‐of‐field dose will be examined. The impact of treatment energy will be discussed, particularly in the context of neutron production and the biological potency of neutrons. Finally, approaches to measure and calculate the out‐of‐field dose will be examined, and the limitations and challenges of many such approaches will be explored. Learning Objectives: 1. Understand the magnitude of out‐of‐field dose and treatment parameters that affect it 2. Understand the limitations of treatment planning system calculations 3. Understand techniques and challenges of measuring out‐of‐field doses 4. Understand the impact of these doses on patient care

Purpose: After a new in‐vault CT‐on‐rails system repeatedly malfunctioned following use of a high‐energy radiotherapy beam, we investigated the presence and impact of neutronradiation on this electronic system, as well as potential neutron shielding and neutron evaluation options. Methods: We first determined the CT scanner's failure rate as a function of the number of 18 MV monitor units (MUs) delivered. We then reexamined the failure rate with both 2.7‐cm‐thick and 7.6‐cm‐thick borated polyethylene (BPE) covering the linac head for neutron shielding. To further examine shielding options, as well as to explore which neutrons were relevant to the scanner failure, Monte Carlo was used to calculate the neutron fluence and spectrum in the bore of the CTscanner. Simulations included BPE covering the CTscanner itself as well as covering the linac head. Results: We found that the CTscanner had a 57% chance of failure after the delivery of 200 MUs. While the addition of neutron shielding to the accelerator head reduced this risk of failure, the benefit was minimal and even 7.6 cm of BPE was still associated with a 29% chance of failure after the delivery of 200 MU. This shielding benefit was achieved regardless of whether the linac head or CTscanner was shielded. Additionally, it was determined that fast neutrons were primarily responsible for the electronic failures. Conclusions: As illustrated by the CT‐on‐rails system in the current study, physicists should be aware that electronic systems may be highly sensitive to neutronradiation.Physicists should therefore monitor electronic systems that have not been evaluated for potential neutron sensitivity. This is particularly relevant as electronics are increasingly common in the therapy vault and may exhibit increased sensitivity in newer electronic systems.

Purpose:TLD exhibits a response that is dependent on photon energy; however, the specific energy response is typically determined only for the photon spectrum at the standard dose calibration point (depth of maximum dose on central axis from a 10cm×10cm field). Variations in the energy spectra as a result of field size, measurement location, or the presence of heterogeneities have not been previously considered. Therefore, we sought to quantify these variations from a 6‐MV beam, as well as to characterize the corresponding changes in energy response of thermoluminescent dosimeters(TLD).Method and Materials:Photon energy spectra from a 6MV beam were simulated at various locations throughout a water tank using Monte Carlo. This was done for fields sizes ranging from 5cm×5cm to 20cm×20cm, and for locations up to 20cm deep and 50cm from the central axis. We also calculated the spectra with the addition of 9cm of bone and lung material. The range of calculated energy spectra were then used to determine theoretical energy correction factors of TLD through the application of Burlin cavity theory. Results: The photon energy spectrum varied with location and field size, and was most substantially different outside of the treatment field. The presence of heterogeneities had little impact on the spectrum. Correspondingly, within the treatment field, as well as in the presence of heterogeneities, there was only a small perturbation to the dose measurement of the TLD (

Purpose: To compare a custom‐developed method for accurate dose recalculation of patient plans entered into clinical trials with results from a common treatment planning system. Method and Materials: A measurement‐driven multiple‐source model with the Dose Planning Method (DPM) Monte Carlo(MC)dose calculation algorithm was previously developed, validated, and benchmarked for the Varian 6 MV and 10 MV photon beams. Several patient cases have been recalculated and compared to the calculations from a Pinnacle planning system. Intensity modulated radiation therapy(IMRT) prostate, IMRT abdomen, stereotactic body radiotherapylung, and IMRTlung patient cases were selected. Results:Field sizes from 4 cm × 4 cm to 40 cm × 40 cm were validated to within 2% of the maximum dose and 2 mm distance to agreement. At least 95% of the data tested met the validation criteria. Benchmark treatments planned using anthropomorphic phantoms were tested to within 3% of the target dose and 2 mm distance to agreement. At least 85% of the data tested met the benchmark criteria. Disagreement in the patient plan evaluation tended to occur at heterogeneity interfaces where electronic disequilibrium occurred, and in the beam penumbra, where scattered radiation was more prominent. The ratios of planning system calculation to MC calculation for the mean dose of the gross and planning target volumes for the patient plans ranged from 0.984 to 1.016. Conclusion: These results show that this MCsoftware code generates answers similar to the Pinnacle system for IMRT and SBRT treatment plans. Differences are consistent with the superior physics modeling inherent in the Monte Carlo code. We believe the method will be useful for recalculating dose distributions for patients entered into clinical trials. Work supported by PHS CA010953, CA081647, and R01 CA85181 awarded by NCI, DHHS

Purpose: Compare the out‐of‐field secondary doses to critical organs in pediatric patients resulting from three different craniospinal irradiation (CSI) treatments. Methods and Materials: A pediatric anthropomorphic Rando phantom was CT scanned for planning purposes following MDACC protocols. Based on the images three different treatment plans employing photon, electron and proton spinal fields, were designed. A common dose prescription for pediatric patients with CNS disease of 36 Gy to the cerebrospinal axis was used for all three plans. The photon and electron spinal fields plans were each delivered three times and the doses at several critical organ sites were measured with TLD. The proton delivery was simulated with MCNPX to calculate the out‐of‐field secondary neutron contribution to the doses at the same critical organ sites. Measured doses for all three treatments were compared. Results: In the majority of the investigated critical organ sites, the doses from the photon spinal fields were higher than from the electron spinal fields. Many of the critical organ doses beyond the spinal target were greater for the photon delivery than for the electron delivery by several orders of magnitude. The entrance doses, as well as the doses lateral to and outside of the target, were up to a factor of two greater for the electron delivery. The expectation for the proton delivery is that the out‐of‐field doses from the secondary neutrons will be less than the conventional electron and photon treatments. Conclusions: An improved clinical treatment decision for pediatric patients can be made using the comparisons between the out‐of field doses for three different treatment methods. Work supported by CA010953 awarded by NCI, DHHS

Purpose: The purpose of this work was to quantify the discrepancy between the out‐of‐field doses as reported by a treatment planning system (TPS) to the actual measured doses.Method and Materials: Out‐of‐field doses were determined at the same location in phantom for a range of distances from the field edge using the treatment planning system (TPS) and thermoluminescent dosimeters(TLD). A typical radiation plan for Hodgkin lymphoma was created using Eclipse TPS (Varian Medical Systems, Palo Alto, CA) for an anthropomorphic male dosimetry phantom. TLD capsules were placed in the phantom at distances ranging from 1.25 cm to 18.75 cm inferior to the field edge. The phantom was irradiated with the full treatment plan dose.TLDs were read using an established laboratory protocol with an error of ⩽3%. For comparison, the out‐of‐field doses at the same positions were determined in the TPS using the point dose measurement tool. Results: As expected, the doses from both the TLD and TPS decrease exponentially as a function of distance from the field edge. Close to the field edge (within 1.25 cm), the TPS overestimated the actual dose by approximately 13%. At larger distances from the field edge the TPS underestimated dose with the magnitude of the underestimation increasing with increasing distance from the field edge. At a distance of 11.25 cm from the field edge, the TPS underestimated the dose by 53%. The TPS reported no dose at distances greater than 12 cm from the field edge. Conclusion: Studies requiring doses to organs out of the treatment field should not rely on the treatment planning system. Measurements using anthropomorphic phantoms or mathematical calculation models that are benchmarked for out‐of‐field doses should be used to determine doses to peripheral organs and tissues.

Purpose: Accurate dosimetry calculations are an important component of epidemiological studies of radiation induced late effects. It is impossible to perform individual dosimetry calculations for thousands of patients. Thus, patients are grouped using several criteria, and dose is calculated for a representative individual from each group. Within groups, the organs are assumed to be in the same location; however, this is complicated when the position of the organ of interest is known to vary between patients, as is the case for the stomach. The purpose of this work was to determine the impact of the size and position of the stomach on the mean organdose when the stomach is completely outside of the treatment field. This work focused on the dose received from a typical mantle field for Hodgkin lymphoma. Method and Materials: A treatment plan was created for an anthropomorphic male dosimetry phantom. Contouring tools in the treatment planning system were used to create five additional stomachs by modifying the size and position of the original stomach contour. The phantom was loaded with 550 thermoluminescent dosimeters(TLD) to encompass the six different stomachs. The phantom was irradiated with the treatment plan and the dosimeters were read using an established laboratory protocol. The mean dose to each of the six stomachs was determined using the appropriate TLD data points. An Analysis of Variance (ANOVA) was performed to determine whether the mean doses to each of the six samples were statistically different. Results: The mean out‐of‐field dose was determined for six common variations of the stomach. The ANOVA indicated that the means of the six stomachs were not statistically different. Conclusion: This study indicates that variations in organ position and size do not statistically change the mean organdose for out‐ of‐field organs.

Purpose:IMRT has become a standard technique for the treatment of many tumors. It is often important to know the out‐of‐field dose associated with such treatments, such as for fetal dose evaluation, or estimating organ doses during epidemiologic study of second cancer risk. While evaluating such doses is routine for conventional therapies (e.g., Task Group 36 data), no simple technique is available to estimate doses from IMRT treatments. This work develops and evaluates such a technique. Method and Materials: A simple equation was developed to calculate the out‐of‐field dose (Dimrt) from an IMRT treatment based on treatment parameters and the out‐of‐field dose from conventional therapy (Dconv) with the same jaw setting (which is readily estimated from such sources as TG‐36). The accuracy of this equation was tested by comparing out‐of‐field doses calculated with the developed equation to those calculated for IMRT treatments using Monte Carlo. Results: The following equation was developed to estimate the out‐of‐field dose from IMRT: Dimrt=Dcon[P×CF + (L+C)×MF], where P, L, and C are the fraction of the out‐of‐field dose originating from patient scatter, head leakage, and collimator scatter respectively, CF is the fraction of the open field covered by the CIAO field (continuous irradiated area opening), and MF is the IMRT MU modulation factor. These values are tabulated for typical treatments. Compared to the out‐of‐field dose from IMRT, doses calculated with this equation generally agreed within 25%. Larger variations were sometimes observed, particularly within 5cm of the edge of the treatment field where large variations in the out‐of‐field dose are expected due to its dependence on the particular treatment plan and planning objectives. Conclusion: A simple and suitably accurate equation was derived to calculate the out‐of‐field dose from IMRT treatments. This technique represents a viable approach for estimating the out‐of‐field dose for clinical or epidemiological applications.

Purpose: Head leakage, collimatorscatter, and patient scatter all contribute to dose outside of the treatment field during radiation therapy, which may lead to sequelae such as second cancers. In order to minimize out‐of‐field dose, it is essential to quantify the sources of this undesired radiation, particularly in the era of evolving radiation therapy.Method and Materials: The out‐of‐field dose was calculated with a previously benchmarked model of a Varian 2100 linac that was equipped with a 120 leaf MLC and was developed in MCNPX. The total out‐of‐field dose was calculated as a function of distance from the field edge for different field sizes, treatment energy, and linac configurations. In particular, the impact of removing the MLC, removing the flattening filter, and modulating the treatment field (IMRT) were evaluated. For each field and linac configuration, the head leakage, collimatorscatter, and patient scatter components were resolved by repeating simulations but eliminating particles either leaving the MLC opening (thereby isolating for head leakage), or entering the phantom within the primary field (thereby isolating for head leakage plus collimatorscatter).Results: In general, patient scatter dominated within ∼20 cm of the treatment field, while head leakage dominated at distances beyond ∼20 cm from the field edge. Collimatorscatter typically made up 20% of the out‐of‐field dose, regardless of distance from the field edge. Removal of the MLC increased head leakage, while removal of the flattening filter decreased head leakage and collimatorscatter, but increased patient scatter. Field modulation increased head leakage and collimatorscatter, while patient scatter was unchanged. Conclusion: Different linear accelerator configurations changed the relative importance of head leakage, collimatorscatter, and patient scatter contributions to the out‐of‐field dose. These differences should be considered when evaluating the out‐of‐field dose and techniques to reduce the out‐of‐field dose.

Purpose: Evaluate the performance of an implantable MOSFETdosimeter designed for hypo‐fractionated protocols. Method and Materials: The DVS® implantable dosimeter has been developed for use with external beam radiotherapy. A new version is currently under development for use with hypo‐fractionated protocols (DVS‐HFT*). This dosimeter is pre‐calibrated for use at body temperature in a dose per fraction range of 340–1200 cGy. The dosimeter response was evaluated in a water phantom at 37°C to validate the accuracy as a function of dose. Since delivery times for high dose fractions can vary, a study of the response as a function of treatment time was performed. In addition, dose rate dependency for high doses was evaluated. Final dosimeter accuracy was validated by delivering simulated hypo‐fractionated treatments on a body phantom at 37 °C. Two dosimeters each (total six) were irradiated inside the phantom with a breast BID plan at 340 cGy per fraction (10 fx), two prostate IMRT plans at 700 cGy per fraction (5 fx) using 6 MV and 18 MV. Results: The dosimeter had minimum dependency on dose fraction size from 340 to 1200 cGy. Phantom testing found a response accuracy of

Purpose: This work measures peripheral doses and estimates the risk of secondary fatal malignancies associated with pediatric brain and adult prostate treatments delivered with a Cyberknife and a conventional, gantry‐based accelerator. Method and Materials: Comparable Cyberknife and IMRTbraintreatment plans were developed for a pediatric anthropomorphic Rando phantom using the Cyberknife Multiplan and Phillips Pinnacle treatment planning system, respectively. Similarly, adult prostate treatment plans were generated using the Multiplan and Pinnacle planning systems for an adult Rando phantom. Target and organs at risk were contoured for each phantom and equivalent, clinically appropriate normal tissue constraints were utilized for each treatment modality. TLD were positioned in specific organ locations within each anthropomorphic phantom and were irradiated three times per plan. The average dose was determined for each out‐of‐field organ site and was compared between the two different treatment delivery devices. Organ weighted, linear non‐threshold dose response model risk factors were then used to estimate the risk of secondary fatal malignancies.Results:Doses calculated from the adult TLD data were equal to or lower for all organs with the IMRTtreatment, and hence, the overall risk was lower. The pediatric TLDdose findings were mixed between the IMRT and Cyberknife treatments.TLD located in cranial organ sites, proximal to the treatment field, exhibited higher readings for the IMRTtreatment. However, the Cyberknife treatment consistently resulted in higher average doses at all distant organ sites. Conclusion: The overall risk of secondary fatal malignancies was higher for the Cyberknife compared to the IMRT adult and pediatric treatments, with the exception of critical organs adjacent to the treatment field for the pediatric case. This appears to be due to higher out‐of‐field secondary radiation doses resulting from the greater number of monitor units and longer treatment times associated with Cyberknife treatments.

Purpose: To evaluate out‐of‐field dose from a medical accelerator with the flattening‐filter removed as compared to with it present. Method and Materials: A previously developed Monte Carlomodel of a Varian 2100 accelerator was used to calculate the out‐of‐field photon dose with and without the flattening filter present. Calculation were done for static fields of 5×5, 10×10, and 15×15 cm2 at 6 and 18 MV incident on a water tank. Additional calculations were conducted for IMRT treatments of the prostate at 6 and 18 MV and for a pediatric brain lesion at 6 MV. To compare the relative out‐of‐field dose from the IMRT treatments, the risk of secondary malignancy in the low dose region was calculated for each. Results: For both static fields and IMRT treatments, the out‐of‐field doses were comparable or higher with the filter removed than with it present at locations near the treatment field. At locations farther from the treatment field, the out‐of‐field doses were substantially lower without the filter. More reduction in the out‐of‐field dose was observed at 18 MV than at 6 MV. For the clinical treatments, as evaluated by the risk of secondary malignancies, there was an overall reduction in out‐of‐field photon dose of 10–30% when the flattening filter was removed. Conclusion: The reduction in out‐of‐field dose from removal of the flattening filter depends on the clinical case being examined, including the target volume and the location of sensitive organs. Treatments involving large target volumes near sensitive organs are likely to produce as much, or more, out‐of‐field dose when the flattening filter is removed. However, for several common clinical scenarios, such as IMRT for the prostate or a small pediatric brain lesion, an overall reduction in out‐of‐field dose was found on the order of 10–30%. Conflict of Interest: Supported by Varian Medical Systems.

Purpose: To evaluate the potential for an implantable dosimeter to be also used as an internal fiducial maker. Additionally, to evaluate the synergy of the dosimeter/fiducial capabilities. Method and Materials: Two implantable MOSFETdetectors (DVS®, Sicel Technologies, Inc.) were placed inside an acrylic pelvic phantom for which a wedged‐field photon plan and an eight‐field prostate treatment plan were developed. For each plan, conditions were simulated so that detectors were in their correct positions or slightly displaced to represent patient setup error and/or organ motion. Doses measured by the two detectors after irradiation were compared to those calculated by the treatment planning software. Additionally, using localization software (MarkerVision, Varian MedicalSystems) and kilovoltage images of each setup, the displacement of the detectors from their correct locations was calculated and compared to the induced physical displacement. Results: For all alignments and detector positions, measured and calculated doses showed an average disagreement of 2.7%. The detectors were easily visualized in AP and lateral radiographs, and the induced detector displacements were typically recognized by the localization software within 0.1 cm, and were recognized within 0.16 cm at worst. The relationship between the geometric displacement and measured dose was dependent on the positioning of the dosimeter and dose gradients. If the detector was placed in a uniform dose region then geometric displacements did not induce any dosimetric discrepancies. Alternately, if the detector was placed in a strong dose gradient then even high geometric accuracy resulted in dosimetric discrepancies. Conclusion: The implantable detector functioned well as both an internal dosimeter and as an internal fiducial marker and thus may be useful as a clinical tool to localize the target volume and verify dose delivery in vivo. Conflict of Interest: Research was supported by a grant from Varian MedicalSystems.