Cory Knill

Monte Carlo calculated correction factors for the PTW microDiamond detector in the Gamma Knife-Model C.

PURPOSE Accurate dose measurements in small fields require correction factors when sufficient CPE is not present inside of the field. These factors adjust for perturbation, volume averaging, and other effects; as such, they are field size, detector, and phantom dependent. In this work, Monte Carlo (MC) methods were used to calculate correction factors for PTWs microDiamond detector in Elektas Gamma Knife Model-C unit. These correction factors allow for accurate measurement of output factors-even in the smallest field sizes where CPE is not present. METHODS The small field correction factors were calculated as kQclin,Qmsr (fclin,fmsr) correction factors according to the Alfonso formalism. The MC model of the Gamma Knife was built with the EGSnrc code system, using BEAMnrc and DOSRZnrc user codes. Efforts were made to validate the MC model against experimental measurements. Using the model, field output factors and measurement ratios for each of the four helmet sizes were simulated for an ABS plastic phantom and validated against film measurements, detector measurements, and treatment planning system (TPS) data. Once validated against the available ABS phantom, the model was applied to a more waterlike solid water phantom. Using MC results from the solid water phantom, the final k correction factors were determined relative to the machine specific reference field-the 18 mm helmet, which is the largest field size available on the unit. RESULTS When validating against experimental measurements using the ABS phantom, all MC methods agreed with experiment within the stated uncertainties: MC determined field output factors agreed within 0.6% of the TPS and 1.4% of film; and MC simulated measurement ratios matched physically measured ratios within 1% for all helmet sizes. kQclin,Qmsr (fclin,fmsr) for the PTW microDiamond in the solid water phantom approached unity to within 0.4% ± 1.7% for all the helmet sizes except the 4 mm; the 4 mm helmet size over-responded by 3.2% ± 1.7%, resulting in a kQ4mm,Q18mm (f4mm,f18mm) of 0.969. CONCLUSIONS Similar to what has been found in the Gamma Knife Perfexion, the PTW microDiamond over-responds in the smallest 4 mm field. The over-response can be corrected via the Alfonso formalism using the correction factors determined in this work. Using the MC calculated correction factors, the PTW microDiamond detector is an effective dosimeter in all available helmet sizes.

An analysis of confidence limit calculations used in AAPM Task Group No. 119

PURPOSE The report issued by AAPM Task Group No. 119 outlined a procedure for evaluating the effectiveness of IMRT commissioning. The procedure involves measuring gamma pass-rate indices for IMRT plans of standard phantoms and determining if the results fall within a confidence limit set by assuming normally distributed data. As stated in the TG report, the assumption of normally distributed gamma pass rates is a convenient approximation for commissioning purposes, but may not accurately describe the data. Here the authors attempt to better describe gamma pass-rate data by fitting it to different distributions. The authors then calculate updated confidence limits using those distributions and compare them to those derived using TG No. 119 method. METHODS Gamma pass-rate data from 111 head and neck patients are fitted using the TG No. 119 normal distribution, a truncated normal distribution, and a Weibull distribution. Confidence limits to 95% are calculated for each and compared. A more general analysis of the expected differences between the TG No. 119 method of determining confidence limits and a more time-consuming curve fitting method is performed. RESULTS The TG No. 119 standard normal distribution does not fit the measured data. However, due to the small range of measured data points, the inaccuracy of the fit has only a small effect on the final value of the confidence limits. The confidence limits for the 111 patient plans are within 0.1% of each other for all distributions. The maximum expected difference in confidence limits, calculated using TG No. 119s approximation and a truncated distribution, is 1.2%. CONCLUSIONS A three-parameter Weibull probability distribution more accurately fits the clinical gamma index pass-rate data than the normal distribution adopted by TG No. 119. However, the sensitivity of the confidence limit on distribution fit is low outside of exceptional circumstances.

Investigating ion recombination effects in a liquid‐filled ionization chamber array used for IMRT QA measurements

PURPOSE PTWs Octavius 1000 SRS array performs IMRT quality assurance (QA) measurements with liquid-filled ionization chambers (LICs) to allow closer detector spacing and higher resolution, compared to air-filled QA devices. However, reduced ion mobility in LICs relative to air leads to increased ion recombination effects and reduced collection efficiencies that are dependent on Linac pulse frequency and pulse dose. These pulse parameters are variable during an IMRT delivery, which affects QA results. In this study, (1) 1000 SRS collection efficiencies were measured as a function of pulse frequency and pulse dose, (2) two methods were developed to correct changes in collection efficiencies during IMRT QA measurements, and the effects of these corrections on QA pass rates were compared. METHODS To obtain collection efficiencies, the OCTAVIUS 1000 SRS was used to measure open fields of varying pulse frequency, pulse dose, and beam energy with results normalized to air-filled chamber measurements. Changes in ratios of 1000 SRS to chamber measured dose were attributed to changing collection efficiencies, which were then correlated to pulse parameters using regression analysis. The usefulness of the derived corrections was then evaluated using 6 MV and 10FFF SBRT RapidArc plans delivered to the OCTAVIUS 4D system using a TrueBeam (Varian Medical Systems) linear accelerator equipped with a high definition multileaf collimator. For the first correction, matlab software was developed that calculates pulse frequency and pulse dose for each detector, using measurement and DICOM RT Plan files. Pulse information is converted to collection efficiency, and measurements are corrected by multiplying detector dose by ratios of calibration to measured collection efficiencies. For the second correction the MU/min in the daily 1000 SRS calibration was chosen to match the average MU/min of the volumetric modulated arc therapy plan. Effects of the two corrections on QA results were examined by performing 3D gamma analysis comparing predicted to measured dose, with and without corrections. RESULTS Collection efficiencies correlated linearly to pulse dose, while correlations with pulse frequency were less defined, generally increasing as pulse frequency decreased. After complex matlab corrections, average 3D gamma pass rates improved by [0.07%,0.40%,1.17%] for 6 MV and [0.29%,1.40%,4.57%] for 10FFF using [3%/3 mm,2%/2 mm,1%/1 mm] criteria. Maximum changes in gamma pass rates were [0.43%,1.63%,3.05%] for 6 MV and [1.00%,4.80%,11.2%] for 10FFF using [3%/3 mm,2%/2 mm,1%/1 mm] criteria. On average, pass rates of simple daily calibration corrections were within 1% of complex matlab corrections. CONCLUSIONS OCTAVIUS 1000 SRS ion recombination effects have little effect on 6 MV measurements. However, the effect could potentially be clinically significant for higher pulse dose unflattened beams when using tighter gamma tolerances, especially when small aperture sizes are used, as is common for SRS/SBRT. In addition, ion recombination effects are strongly correlated to changing MU/min, therefore MU/min used in daily 1000 SRS calibrations should be matched to the expected average MU/min of the IMRT plan.

Fast Monte Carlo simulation for total body irradiation using a 60Co teletherapy unit

Our institution delivers TBI using a modified Theratron 780  60Co unit. Due to limitations of our treatment planning system in calculating dose for this treatment, we have developed a fast Monte Carlo code to calculate dose distributions within the patient. The algorithm is written in C and uses voxel density information from CT images to calculate dose in heterogeneous media. To test the algorithm, film‐based dose measurements were made separately in a simple water phantom with a high‐density insert and a RANDO phantom and then compared to doses calculated by the Monte Carlo algorithm. In addition, a separate simulation in GEANT4 was run for the RANDO phantom and compared to both film and the in‐house simulation. All results were analyzed using RIT113 film analysis software. Simulations in the water phantom accurately predict the depth of maximum dose in the phantom at 0.5 cm. The measured PDD along the central axis of the beam closely matches the PDD generated from the Monte Carlo code, deviating on average by only 3% along the depth of the water phantom. Dose measured at planes inside the high‐density insert had a mean difference of 4.9% on cross‐profile measurement. In the RANDO phantom, gamma pass rates vary between 91% and 99% at 3 mm, 3%, and were >99% at 5 mm, 5% for the four film planes measured. Profiles taken across the film and both simulations resulted in mean relative differences of <2% for all profiles in each slice measured. The Monte Carlo algorithm presented here is potentially a viable method for calculating dose distributions delivered in TBI treatments at our center. While not yet refined enough to be the primary method of treatment planning, the algorithm at its current resolution determines the dose distribution for one patient within a few hours, and provides clinically useful information in planning TBI. With appropriate optimization, the Monte Carlo method presented here could potentially be implemented as a first‐line treatment planning option for  60Co TBI. PACS number: 87.10.Rt

Modeling the Agility MLC in the Monaco treatment planning system

We investigate the relationship between the various parameters in the Monaco MLC model and dose calculation accuracy for an Elekta Agility MLC. The vendor‐provided MLC modeling procedure — completed first with external vendor participation and then exclusively in‐house — was used in combination with our own procedures to investigate several sets of MLC modeling parameters to determine their effect on dose distributions and point‐dose measurements. Simple plans provided in the vendor procedure were used to elucidate specific mechanical characteristics of the MLC, while ten complex treatment plans — five IMRT and five VMAT — created using TG‐119‐based structure sets were used to test clinical dosimetric effects of particular parameter choices. EDR2 film was used for the vendor fields to give high spatial resolution, while a combination of MapCHECK and ion chambers were used for the in‐house TG‐119‐based procedures. The vendor‐determined parameter set provided a reasonable starting point for the MLC model and largely delivered acceptable gamma pass rates for clinical plans — including a passing external evaluation using the IROC H&N phantom. However, the vendor model did not provide point‐dose accuracy consistent with that seen in other treatment systems at our center. Through further internal testing it was found that there existed many sets of MLC parameters, often at opposite ends of their allowable ranges, that provided similar dosimetric characteristics and good agreement with planar and point‐dose measurements. In particular, the leaf offset and tip leakage parameters compensated for one another if adjusted in opposite directions, which provided a level curve of acceptable parameter sets across all plans. Interestingly, gamma pass rates of the plans were less dependent upon parameter choices than point‐dose measurements, suggesting that MLC modeling using only gamma evaluation may be generally an insufficient approach. It was also found that exploring all parameters of the very robust MLC model to find the best match to the vendor‐provided QA fields can reduce the pass rates of the TG‐119‐based clinical distributions as compared to simpler models. A wide variety of parameter sets produced MLC models capable of meeting RPC passing criteria for their H&N IMRT phantom. The most accurate models were achievable using a combination of vendor‐provided and in‐house procedures. The potential existed for an over‐modeling of the Agility MLC in an effort to obtain the fine structure of certain quality assurance fields, which led to a reduction in agreement between calculation and measurement of more typical clinical dose distributions. PACS number(s): 87.56.nk, 87.53.Kn, 87.55.km, 87.55.QrWe investigate the relationship between the various parameters in the Monaco MLC model and dose calculation accuracy for an Elekta Agility MLC. The vendor-provided MLC modeling procedure - completed first with external vendor participation and then exclusively in-house - was used in combination with our own procedures to investigate several sets of MLC modeling parameters to determine their effect on dose distributions and point-dose measurements. Simple plans provided in the vendor procedure were used to elucidate specific mechanical characteristics of the MLC, while ten complex treatment plans - five IMRT and five VMAT - created using TG-119-based structure sets were used to test clinical dosimetric effects of particular parameter choices. EDR2 film was used for the vendor fields to give high spatial resolution, while a combination of MapCHECK and ion chambers were used for the in-house TG-119-based procedures. The vendor-determined parameter set provided a reasonable starting point for the MLC model and largely delivered acceptable gamma pass rates for clinical plans - including a passing external evaluation using the IROC H&N phantom. However, the vendor model did not provide point-dose accuracy consistent with that seen in other treatment systems at our center. Through further internal testing it was found that there existed many sets of MLC parameters, often at opposite ends of their allowable ranges, that provided similar dosimetric characteristics and good agreement with planar and point-dose measurements. In particular, the leaf offset and tip leakage parameters compensated for one another if adjusted in opposite directions, which provided a level curve of acceptable parameter sets across all plans. Interestingly, gamma pass rates of the plans were less dependent upon parameter choices than point-dose measurements, suggesting that MLC modeling using only gamma evaluation may be generally an insufficient approach. It was also found that exploring all parameters of the very robust MLC model to find the best match to the vendor-provided QA fields can reduce the pass rates of the TG-119-based clinical distributions as compared to simpler models. A wide variety of parameter sets produced MLC models capable of meeting RPC passing criteria for their H&N IMRT phantom. The most accurate models were achievable using a combination of vendor-provided and in-house procedures. The potential existed for an over-modeling of the Agility MLC in an effort to obtain the fine structure of certain quality assurance fields, which led to a reduction in agreement between calculation and measurement of more typical clinical dose distributions. PACS number(s): 87.56.nk, 87.53.Kn, 87.55.km, 87.55.Qr.

Development and evaluation of an end-to-end test for head and neck IMRT with a novel multiple-dosimetric modality phantom

A comprehensive end‐to‐end test for head and neck IMRT treatments was developed using a custom phantom designed to utilize multiple dosimetry devices. Initial end‐to‐end test and custom H&N phantom were designed to yield maximum information in anatomical regions significant to H&N plans with respect to: (i) geometric accuracy, (ii) dosimetric accuracy, and (iii) treatment reproducibility. The phantom was designed in collaboration with Integrated Medical Technologies. The phantom was imaged on a CT simulator and the CT was reconstructed with 1 mm slice thickness and imported into Varians Eclipse treatment planning system. OARs and the PTV were contoured with the aid of Smart Segmentation. A clinical template was used to create an eight‐field IMRT plan and dose was calculated with heterogeneity correction on. Plans were delivered with a TrueBeam equipped with a high definition MLC. Preliminary end‐to‐end results were measured using film, ion chambers, and optically stimulated luminescent dosimeters (OSLDs). Ion chamber dose measurements were compared to the treatment planning system. Films were analyzed with FilmQA Pro using composite gamma index. OSLDs were read with a MicroStar reader using a custom calibration curve. Final phantom design incorporated two axial and one coronal film planes with 18 OSLD locations adjacent to those planes as well as four locations for IMRT ionization chambers below inferior film plane. The end‐to‐end test was consistently reproducible, resulting in average gamma pass rate greater than 99% using 3%/3 mm analysis criteria, and average OSLD and ion chamber measurements within 1% of planned dose. After initial calibration of OSLD and film systems, the end‐to‐end test provides next‐day results, allowing for integration in routine clinical QA. Preliminary trials have demonstrated that our end‐to‐end is a reproducible QA tool that enables the ongoing evaluation of dosimetric and geometric accuracy of clinical head and neck treatments. PACS number(s): 87.55.Qr

Multi-institutional evaluation of end-to-end protocol for IMRT/VMAT treatment chains utilizing conventional linacs

We conducted a multi-institutional assessment of a recently developed end-to-end monthly quality assurance (QA) protocol for external beam radiation therapy treatment chains. This protocol validates the entire treatment chain against a baseline to detect the presence of complex errors not easily found in standard component-based QA methods. Participating physicists from 3 institutions ran the end-to-end protocol on treatment chains that include Imaging and Radiation Oncology Core (IROC)-credentialed linacs. Results were analyzed in the form of American Association of Physicists in Medicine (AAPM) Task Group (TG)-119 so that they may be referenced by future test participants. Optically stimulated luminescent dosimeter (OSLD), EBT3 radiochromic film, and A1SL ion chamber readings were accumulated across 10 test runs. Confidence limits were calculated to determine where 95% of measurements should fall. From calculated confidence limits, 95% of measurements should be within 5% error for OSLDs, 4% error for ionization chambers, and 4% error for (96% relative gamma pass rate) radiochromic film at 3% agreement/3 mm distance to agreement. Data were separated by institution, model of linac, and treatment protocol (intensity-modulated radiation therapy [IMRT] vs volumetric modulated arc therapy [VMAT]). A total of 97% of OSLDs, 98% of ion chambers, and 93% of films were within the confidence limits; measurements were found outside these limits by a maximum of 4%, < 1%, and < 1%, respectively. Data were consistent despite institutional differences in OSLD reading equipment and radiochromic film calibration techniques. Results from this test may be used by clinics for data comparison. Areas of improvement were identified in the end-to-end protocol that can be implemented in an updated version. The consistency of our data demonstrates the reproducibility and ease-of-use of such tests and suggests a potential role for their use in broad end-to-end QA initiatives.

SU‐F‐T‐584: Investigating Correction Methods for Ion Recombination Effects in OCTAVIUS 1000 SRS Measurements

PURPOSE PTWs Octavius 1000 SRS array performs IMRT QA measurements with liquid filled ionization chambers (LICs). Collection efficiencies of LICs have been shown to change during IMRT delivery as a function of LINAC pulse frequency and pulse dose, which affects QA results. In this study, two methods were developed to correct changes in collection efficiencies during IMRT QA measurements, and the effects of these corrections on QA pass rates were compared. METHODS For the first correction, Matlab software was developed that calculates pulse frequency and pulse dose for each detector, using measurement and DICOM RT Plan files. Pulse information is converted to collection efficiency and measurements are corrected by multiplying detector dose by ratios of calibration to measured collection efficiencies. For the second correction, MU/min in daily 1000 SRS calibration was chosen to match average MU/min of the VMAT plan. Usefulness of derived corrections were evaluated using 6MV and 10FFF SBRT RapidArc plans delivered to the OCTAVIUS 4D system using a TrueBeam equipped with an HD- MLC. Effects of the two corrections on QA results were examined by performing 3D gamma analysis comparing predicted to measured dose, with and without corrections. RESULTS After complex Matlab corrections, average 3D gamma pass rates improved by [0.07%,0.40%,1.17%] for 6MV and [0.29%,1.40%,4.57%] for 10FFF using [3%/3mm,2%/2mm,1%/1mm] criteria. Maximum changes in gamma pass rates were [0.43%,1.63%,3.05%] for 6MV and [1.00%,4.80%,11.2%] for 10FFF using [3%/3mm,2%/2mm,1%/1mm] criteria. On average, pass rates of simple daily calibration corrections were within 1% of complex Matlab corrections. CONCLUSION Ion recombination effects can potentially be clinically significant for OCTAVIUS 1000 SRS measurements, especially for higher pulse dose unflattened beams when using tighter gamma tolerances. Matching daily 1000 SRS calibration MU/min to average planned MU/min is a simple correction that greatly reduces ion recombination effects, improving measurements accuracy and gamma pass rates. This work was supported by PTW.

SU‐E‐T‐485: Investigation of a Synthetic Diamond Detector for Tomotherapy Dosimetry

PURPOSE Tomotherapy treatments are characterized by rotational deliveries of flattening-filter free fields resulting in high-gradient dose distributions. Small volume, rotationally independent detectors are needed for accurate dosimetry. PTWs microDiamond detector, with its small sensitive volume (0.004mm3 ), could potentially be an ideal detector for Tomotherapy. The microDiamond detector was tested against a small volume Exradin A1SL ion chamber for Tomotherapy open-field and IMRT commissioning measurements. METHODS Custom detector holders were fabricated to allow A1SL and microDiamond measurements in the Tomotherapy Cheese phantom and a square solid water phantom. The microDiamond rotational dependence within the Tomotherapy phantom was tested by incrementally rotating the detector in between static-gantry angle Tomotherapy irradiations. Longitudinal Tomotherapy profiles, for all field sizes, were measured with the microDiamond and A1SL detectors at 1.5cm depth in the square phantom, and compared to film. Detector axes were aligned parallel to table motion. Per TG-119 recommendations, both detectors were calibrated to known doses in phantoms and used to measure high-dose points in TG-119 H&N and Prostate plans. The measurements were compared to the treatment planning system and subsequently compared to published TG-119 confidence limits. RESULTS The microDiamond angular dependence was less than 0.5%. The average difference between the detectors and film-measured longitudinal profile 80-20% penumbras were 0.03+/-0.04mm and 1.36+/-0.22mm for the microDiamond and A1SL, respectively. The average difference between the detector and filmmeasured field sizes were 0.07+/-0.01mm and 0.09+/-0.02mm for the microDiamond and A1SL, respectively. The measured confidence limits were 0.023 and 0.015 for microDiamond and A1SL, respectively. TG-119 reported a confidence limit of 0.034. CONCLUSION The microDiamond measured open-field longitudinal Tomotherapy profiles more closely resembled film measurements, compared with the A1Sl chamber. Furthermore, the microDiamond was shown to have low angular dependence and both detectors were able to accurately measure the highdose points in TG-119 H&N and prostate plans. Our institution has research grants with PTW.

SU-E-T-109: Development of An End-To-End Test for the Varian TrueBeamtm with a Novel Multiple-Dosimetric Modality H&N Phantom

PURPOSE To develop a comprehensive end-to-end test for Varians TrueBeam linear accelerator for head and neck IMRT using a custom phantom designed to utilize multiple dosimetry devices. METHODS The initial end-to-end test and custom H&N phantom were designed to yield maximum information in anatomical regions significant to H&N plans with respect to: i) geometric accuracy, ii) dosimetric accuracy, and iii) treatment reproducibility. The phantom was designed in collaboration with Integrated Medical Technologies. A CT image was taken with a 1mm slice thickness. The CT was imported into Varians Eclipse treatment planning system, where OARs and the PTV were contoured. A clinical template was used to create an eight field static gantry angle IMRT plan. After optimization, dose was calculated using the Analytic Anisotropic Algorithm with inhomogeneity correction. Plans were delivered with a TrueBeam equipped with a high definition MLC. Preliminary end-to-end results were measured using film and ion chambers. Ion chamber dose measurements were compared to the TPS. Films were analyzed with FilmQAPro using composite gamma index. RESULTS Film analysis for the initial end-to-end plan with a geometrically simple PTV showed average gamma pass rates >99% with a passing criterion of 3% / 3mm. Film analysis of a plan with a more realistic, ie. complex, PTV yielded pass rates >99% in clinically important regions containing the PTV, spinal cord and parotid glands. Ion chamber measurements were on average within 1.21% of calculated dose for both plans. CONCLUSION trials have demonstrated that our end-to-end testing methods provide baseline values for the dosimetric and geometric accuracy of Varians TrueBeam system.