Are simple IMRT beams more robust against MLC error? Exploring the impact of MLC errors on planar quality assurance and plan quality for different complexity beams

Evaluation of plan quality and treatment efficiency in cranial stereotactic radiosurgery treatment plans with a variable source-to-axis distance

INTRODUCTION

Radiotherapy deliveries with dynamic couch motions that shorten the source-to-axis distance (SAD) on a C-arm linac have the potential to increase treatment efficiency through the increase of the effective dose rate. In this investigation, we convert clinically deliverable volumetric modulated arc therapy (VMAT) and dynamic conformal arc (DCA) plans for cranial radiosurgery into virtual isocenter plans through implementation of couch trajectories that maintain the target at a shortened but variable SAD throughout treatment.

MATERIALS AND METHODS

A randomly sampled population of patients treated with cranial radiosurgery from within the last three years were separated into groups with one, two, and three lesions. All plans had a single isocenter (regardless of number of targets), and a single prescription dose. Patient treatment plans were converted from their original delivery at a standard isocenter to a dynamic virtual isocenter in MATLAB. The virtual isocenter plan featured a variable isocenter position based upon the closest achievable source-to-target distance (referred to herein as a virtual source-to-axis distance [vSAD]) which avoided collision zones on a TrueBeam STx platform. Apertures were magnified according to the vSAD and monitor units at a given control point were scaled based on the inverse square law. Doses were calculated for the plans with a virtual isocenter in the Eclipse (v13.6.23) treatment planning system (TPS) and were compared with the clinical plans. Plan metrics (MU, Paddick conformity index, gradient index, and the volume receiving 12 Gy or more), normal brain dose-volume differences, as well as maximum doses received by organs at risk (OARs) were assessed. The values were compared between standard and virtual isocenter plans with Wilcoxon Sign Ranked tests to determine significance. A subset of the plans were mapped to the MAX-HD anthropomorphic phantom which contained an insert housing EBT3 GafChromic film and a PTW 31010 microion chamber for dose verification on a linac.

RESULTS

Delivering plans at a virtual isocenter resulted in an average reduction of 20.9% (p = 3×10^-6 ) and 20.6% (p = 3.0×10^-6 ) of MUs across all VMAT and all DCA plans, respectively. There was no significant change in OAR max doses received by plans delivered at a virtual isocenter. The low dose wash (1.0-2.0 Gy or 5-11% of the prescription dose) was increased (by approximately 20 cc) for plans with three lesions. This was equivalent to a 2.7%-3.8% volumetric increase in normal tissue receiving the respective dose level when comparing the plan with a virtual isocenter to a plan with a standard isocenter. Gamma pass rates with a 5%/1mm analysis criteria were 96.40% ± 2.90% and 95.07% ± 3.10% for deliveries at standard and virtual isocenter, respectively. Absolute point dose agreements were within -0.36% ± 3.45% and -0.55% ± 3.39% for deliveries at a standard and virtual isocenter, respectively. Potential time savings per arc were found to have linear relationship with the monitor units delivered per arc (savings of 0.009 s/MU with an r² = 0.866 when fit to plans with a single lesion).

CONCLUSIONS

Converting clinical plans at standard isocenter to a virtual isocenter design did not show any losses to plan quality while simultaneously improving treatment efficiency through MU reductions.

Predictive gamma passing rate of 3D detector array-based volumetric modulated arc therapy quality assurance for prostate cancer via deep learning

To predict the gamma passing rate (GPR) of the three-dimensional (3D) detector array-based volumetric modulated arc therapy (VMAT) quality assurance (QA) for prostate cancer using a convolutional neural network (CNN) with the 3D dose distribution. One hundred thirty-five VMAT plans for prostate cancer were selected: 110 plans were used for training and validation, and 25 plans were used for testing. Verification plans were measured using a helical 3D diode array (ArcCHECK). The dose distribution on the detector element plane of these verification plans was used as input data for the CNN model. The measured GPR (mGPR) values were used as the training data. The CNN model comprises eighteen layers and predicted GPR (pGPR) values. The mGPR and pGPR values were compared, and a cumulative frequency histogram of the prediction error was created to clarify the prediction error tendency. The correlation coefficients of pGPR and mGPR were 0.67, 0.69, 0.66, and 0.73 for 3%/3-mm, 3%/2-mm, 2%/3-mm, and 2%/2-mm gamma criteria, respectively. The respective mean±standard deviations of pGPR-mGPR were -0.87±2.18%, -0.65±2.93%, -0.44±2.53%, and -0.71±3.33%. The probabilities of false positive error cases (pGPR < mGPR) were 72%, 60%, 68%, and 56% for each gamma criterion. We developed a deep learning-based prediction model of the 3D detector array-based VMAT QA for prostate cancer, and evaluated the accuracy and tendency of prediction GPR. This model can provide a proactive estimation for the results of the patient-specific QA before the verification measurement.

intra-Arc binary collimation with dynamic axes trajectory optimization for SRS treatment of multiple metastases with multiple prescriptions

PURPOSE

This work generates multi-metastases cranial SRS/SRT plans using a novel treatment planning technique in which dynamic couch, collimator and gantry trajectories are used with periodic binary target collimation. The performance of this planning technique is evaluated against conventional VMAT planning in terms of various dose and plan quality metrics.

METHODS

A 3D cost space (referred to herein as the combined optimization of dynamic axes, or CODA cube) was calculated based on overlap between targets and OARs and uncollimated areas between targets (island blocking) for all combinations of couch, gantry and collimator angles. Gradient descent through the cube was applied to determine dynamic trajectories. At each control point (CP), each target can either be conformally treated or blocked by the MLC (referred to as intra-arc binary collimation, iABC). Simulated annealing was used to optimize the collimation patterns throughout the arcs as well as the monitor units (MUs) delivered at each CP. Seven previously treated VMAT plans were selected for comparison against the CODA-iABC planning technique. Two CODA-iABC plans were developed: a single gantry arc plan and a plan with one gantry arc and two couch arcs. Plan quality comparison metrics included: maximum and mean dose to organs-at-risk (OARs) (brainstem, chiasm, optic nerves, eyes, and lenses), the volume of normal brain receiving 12Gy (V12Gy), total MUs, target conformity, and dose gradient index.

RESULTS

Treatment plans generated with 1-arc CODA-iABC plans delivered an average of 21% and 30% higher maximum and mean doses to brainstem, respectively, when compared to VMAT plans. Treatment plans generated with 3-arc CODA-iABC used an average of 24% fewer MUs and resulted in an average reduction of 48% maximum dose and 50% mean dose to the OARs, when compared to VMAT. Target conformity values were worse in both CODA-iABC plans than VMAT by an average of 35% and 15%, respectively. There are no significant differences in V12Gy for all three planning techniques, however 3-arc CODA-iABC is more effective at reducing dose to normal brain in the low dose region (<12Gy).

CONCLUSION

CODA-iABC is a novel planning technique that has been developed to automatically generate patient-specific multi-axes trajectories for multiple metastases cranial SRS/SRT. This work has demonstrated the feasibility of planning with this novel method. 1-arc CODA-iABC planning technique is slightly dosimetric inferior to VMAT. With increased sampling of a three-dimensional CODA cube by using a 3-arc CODA-iABC technique, there was improved total dose sparing to all the OARs and increased MU efficiency, but with a cost in target conformity, when compared to VMAT. This article is protected by copyright. All rights reserved.

Evaluation of the ability of three commercially available dosimeters to detect systematic delivery errors in step-and-shoot IMRT plans

Background

There is limited data on error detectability for step-and-shoot intensity modulated radiotherapy (sIMRT) plans, despite significant work on dynamic methods. However, sIMRT treatments have an ongoing role in clinical practice. This study aimed to evaluate variations in the sensitivity of three patient-specific quality assurance (QA) devices to systematic delivery errors in sIMRT plans.

Materials and methods

Four clinical sIMRT plans (prostate and head and neck) were edited to introduce errors in: Multi-Leaf Collimator (MLC) position (increasing field size, leaf pairs offset (1-3 mm) in opposite directions; and field shift, all leaves offset (1-3 mm) in one direction); collimator rotation (1-3 degrees) and gantry rotation (0.5-2 degrees). The total dose for each plan was measured using an ArcCHECK diode array. Each field, excluding those with gantry offsets, was also measured using an Electronic Portal Imager and a MatriXX Evolution 2D ionisation chamber array. 132 plans (858 fields) were delivered, producing 572 measured dose distributions. Measured doses were compared to calculated doses for the no-error plan using Gamma analysis with 3%/3 mm, 3%/2 mm, and 2%/2 mm criteria (1716 analyses).

Results

Generally, pass rates decreased with increasing errors and/or stricter gamma criteria. Pass rate variations with detector and plan type were also observed. For a 3%/3 mm gamma criteria, none of the devices could reliably detect 1 mm MLC position errors or 1 degree collimator rotation errors.

Conclusions

This work has highlighted the need to adapt QA based on treatment plan type and the need for detector specific assessment criteria to detect clinically significant errors.

Evaluation of candidate template beam models for a matched TrueBeam treatment delivery system

PURPOSE

To explore candidate RayStation beam models to serve as a class-specific template for a TrueBeam treatment delivery system.

METHODS

Established validation techniques were used to evaluate three photon beam models: a clinically optimized model from the authors' institution, the built-in RayStation template, and a hybrid consisting of the RayStation template except substituting average MLC parameter values from a recent IROC survey. Comparisons were made for output factors, dose profiles from open fields, as well as representative VMAT test plans.

RESULTS

For jaw-defined output factors, each beam model was within 1.6% of expected published values. Similarly, the majority (57-66%) of jaw-defined dose curves from each model had a gamma pass rate >95% (2% / 3 mm, 20% threshold) when compared to TrueBeam representative beam data. For dose curves from MPPG 5.a MLC-defined fields, average gamma pass rates (1% / 1 mm, 20% threshold) were 92.9%, 85.1%, and 86.0% for the clinical, template, and hybrid models, respectively. For VMAT test plans measured with a diode array detector, median dose differences were 0.6%, 1.3%, and 1.1% for the clinical, template, and hybrid models, respectively. For in-phantom ionization chamber measurements with the same VMAT test plans, the average percent difference was -0.3%, -1.4%, and -1.0% for the clinical, template, and hybrid models, respectively.

CONCLUSION

Beam model templates taken from the vendor and aggregate results within the community were both reasonable starting points, but neither approach was as optimal as a clinically tuned model, the latter producing better agreement with all validation measurements. Given these results, the clinically optimized model represents a better candidate as a consensus template. This can benefit the community by reducing commissioning time and improving dose calculation accuracy for matched TrueBeam treatment delivery systems.

Predictive gamma passing rate by dose uncertainty potential accumulation model

PURPOSE

Intensity-modulated radiation therapy (IMRT) utilizes many small fields for producing a uniform dose distribution. Therefore, there are many field junctions in the target region, and resulting dose uncertainties are accumulated. However, such accumulation of the dose uncertainty has not been implemented in the current practice of IMRT dose verification. The purpose of this study is to develop a method to predict the gamma passing rate (GPR) using a dose uncertainty accumulation model.

METHODS

Thirty-three intensity-modulated (IM) beams for head-and-neck cases with step-and-shoot techniques were used in this study. The treatment plan was created using the XiO treatment planning system (TPS). The IM beam was produced by the ONCOR Impression Plus linear accelerator. MapCHECK was used to measure the dose distribution. The distribution of a dose uncertainty potential (DUP) was generated by in-house software that accumulated field shapes weighted by a segmental monitor unit, followed by Gaussian folding. The width of the Gaussian was determined from the width of the lateral penumbra. The dose difference between the calculated and measured doses was compared with the estimated DUP at each point. The GPR of each beam was predicted for 2%/2-mm, 3%/2-mm, and 3%/3-mm tolerances by its own DUP histogram and a GPR-vs-DUP correlation of other beams using the leave-one-out cross-validation method. The predicted GPR was compared with the measured GPR to evaluate the performance of this prediction method. The criteria for the predicted GPR corresponding to a measured GPR ≥ 90% were estimated to examine the feasibility of estimating the measured GPR by this GPR prediction method.

RESULTS

The DUP was confirmed to have proportionality to the standard deviation (SD) of the dose difference. The SDs of the difference between the measured and predicted GPRs were 3.1, 1.7, and 1.4% for 2%/2-mm, 3%/2-mm, and 3%/3-mm tolerances, respectively. The criteria of the predicted GPR corresponding to the measured GPR ≥ 90% were 94.1 and 95.0% with confidence levels of 99 and 99.9%, respectively.

CONCLUSION

In this study, we confirmed the good proportionality between the dose difference and the estimated DUP. The results showed a feasibility to predict the dose difference from DUP as estimated by a DUP accumulation model. The predicted GPR developed in this study showed good accuracy for planar dose distributions of head and neck IMRT. The prediction method developed in this study is considered to be feasible as a substitute for the current practice of measurement-based verification of the dose distribution with gamma analysis.

Correlation between gamma passing rate and complexity of IMRT plan due to MLC position errors

PURPOSE

This study evaluates the correlation between the susceptibility of the γ passing rate of IMRT plans to the multi-leaf collimator (MLC) position errors and a quantitative plan complexity metric.

METHODS

Twenty patients were selected for this study. For each patient, two IMRT plans were generated using sliding window and step-&-shoot techniques, respectively. Modulation complexity score (MCS) was calculated for all IMRT plans, and symmetric MLC leaf bank errors, ranging from 0.3 mm to 1 mm, were introduced. Original and modified plans were delivered using Varian's Clinac iX. The obtained dose distribution using ArcCHECK was then compared with the TPS calculated dose distribution of the original plans. 3D gamma analysis was performed for each verification with passing criteria of 2%/2 mm. The γ passing rate decreasing gradient were calculated to evaluate relationship between variation of γ passing rate due to MLC errors and complexity.

RESULTS

A linear regression analysis was applied between γ gradient and complexity, and the results showed a linear correlation (R² = 0.81 and 0.82 for open and closed MLC error types, respectively) indicating the more complex plans are more susceptible to MLC leaf bank errors. Meanwhile, correlation of re-normalized γ passing rate and complexity for all errors scenarios also presented a strong correlation (r > 0.75).

CONCLUSION

The statistics results revealed variation relationship of dosimetry robust of plans with various complexities to MLC errors. Our results also suggested that the observed susceptibility is independent of the delivery techniques.

Organ-specific modulation complexity score for the evaluation of dose delivery

The purpose of this study was to correlate the modulation complexity score (MCS) with organ location and to predict potential dose errors for organs before beam delivery for intensity-modulated radiation therapy (IMRT) dosimetry. Sixteen head and neck cancer patients treated with IMRT were selected. Distribution of the relative dose error on each beam was performed using forward projection to the planned dose to compute the predicted dose after doing per-beam quality assurance. Original organ-specific modulation complexity score (oMCS) was created based on a modified MLC, which depended on organ location. First, MCS was calculated based on the change in leaf position between adjacent MLC leaves. Second, the segment edge map (SEM) calculated from the intensity map for each beam was applied to the calculation volume. The oMCS with segment edge (oMCSedge) was derived from the product of oMCS and SEM. The correlation between the dose errors (planned and predicted) and oMCSedge values was evaluated for the target and organs at risk. We have also expanded the original MCS concept to oMCSedge including the organ location. We observed a moderate correlation between the dose errors and oMCSedge for all organs and volumes of interest except the gross tumor volume, brain stem, and spinal cord. In other organs, a moderate improvement in sensitivity was observed on the SEM, which was correlated with dose errors. Although the implementation of oMCSedge would be impractical for normal clinical settings, it is expected that oMCSedge would help a treatment planner to judge whether or not the treatment plan would be acceptably delivered.