Systematic Multileaf Collimator Research Articles

Development of a deep learning-based error detection system without error dose maps in the patient-specific quality assurance of volumetric modulated arc therapy.

To detect errors in patient-specific quality assurance (QA) for volumetric modulated arc therapy (VMAT), we proposed an error detection method based on dose distribution analysis using unsupervised deep learning approach and analyzed 161 prostate VMAT beams measured with a cylindrical detector. For performing error simulation, in addition to error-free dose distribution, dose distributions containing nine types of error, including multileaf collimator (MLC) positional errors, gantry rotation errors, radiation output errors and phantom setup errors, were generated. Only error-free data were employed for the model training, and error-free and error data were employed for the tests. As a deep learning model, the variational autoencoder (VAE) was adopted. The anomaly of test data was quantified by calculating Mahalanobis distance based on the feature vectors acquired from a trained encoder. Based on this anomaly, test data were classified as 'error-free' or 'any-error.' For comparison with conventional approaches, gamma (γ)-analysis was performed, and supervised learning convolutional neural network (S-CNN) was constructed. Receiver operating characteristic curves were obtained to evaluate their performance with the area under the curve (AUC). For all error types, except systematic MLC positional and radiation output errors, the performance of the methods was in the order of S-CNN ˃ VAE-based ˃ γ-analysis (only S-CNN required error data for model training). For example, in random MLC positional error simulation, the AUC of our method, S-CNN and γ-analysis were 0.699, 0.921 and 0.669, respectively. Our results showed that the VAE-based method has the potential to detect errors in patient-specific VMAT QA.

Image-based features in machine learning to identify delivery errors and predict error magnitude for patient-specific IMRT quality assurance

ObjectiveTo identify delivery error type and predict associated error magnitude by image-based features using machine learning (ML).MethodsIn this study, a total of 40 thoracic plans (including 208 beams) were selected, and four error types with different magnitudes were introduced into the original plans, including 1) collimator misalignment (COLL), 2) monitor unit (MU) variation, 3) systematic multileaf collimator misalignment (MLCS), and 4) random MLC misalignment (MLCR). These dose distributions of portal dose predictions for the original plans were defined as the reference dose distributions (RDD), while those for the error-introduced plans were defined as the error-introduced dose distributions (EDD). Both distributions were calculated for all beams with portal dose image prediction (PDIP). Besides, 14 image-based features were extracted from RDD and EDD of portal dose predictions to obtain the feature vectors. In addition, a random forest was adopted for the multiclass classification task, and regression prediction for error magnitude.ResultsThe top five features extracted with the highest weight included 1) the relative displacement in the x direction, 2) the ratio of the absolute minimum residual error to the maximal RDD value, 3) the product of the maximum and minimum residuals, 4) the ratio of the absolute maximum residual error to the maximal RDD value, and 5) the ratio of the absolute mean residual value to the maximal RDD value. The relative displacement in the x direction had the highest weight. The overall accuracy of the five-class classification model was 99.85% for the validation set and 99.30% for the testing set. This model could be applied to the classification of the error-free plan, COLL, MU, MLCS, and MLCR with an accuracy of 100%, 98.4%, 99.9%, 98.0%, and 98.3%, respectively. MLCR had the worst performance in error magnitude prediction (70.1–96.6%), while others had better performance in error magnitude prediction (higher than 93%). In the error magnitude prediction, the mean absolute error (MAE) between predicted error magnitude and actual error ranged from 0.03 to 0.33, with the root mean squared error (RMSE) varying from 0.17 to 0.56 for the validation set. The MAE and RMSE ranged from 0.03 to 0.50 and 0.44 to 0.59 for the test set, respectively.ConclusionIt could be demonstrated in this study that the image-based features extracted from RDD and EDD can be employed to identify different types of delivery errors and accurately predict error magnitude with the assistance of ML techniques. They can be used to associate traditional gamma analysis with clinically based analysis for error classification and magnitude prediction in patient-specific IMRT quality assurance.

Optimizing the Region for Evaluation of Global Gamma Analysis for Nasopharyngeal Cancer (NPC) Pretreatment IMRT QA by COMPASS: A Retrospective Study.

BackgroundThe global gamma passing rate is the most commonly used metric for patient-specific pretreatment quality assurance in radiation therapy. However, the optimal region for evaluation and specific action limits (ALs) need to be explored. Therefore, this study was carried out to explore the optimal region for evaluation of the global gamma passing rate and define ALs by using the COMPASS software.MethodsA total of 93 intensity-modulated radiation therapy (IMRT) plans for nasopharyngeal cancer (NPC) patients, including 61 original plans and 32 multileaf collimator (MLC) error-introduced test plans, were selected for retrospective analysis. Firstly, the dose distribution was divided into six isodose regions (“≥10%”, “≥20%”, “≥30%”, “≥40%”, “≥50%”, and “≥60%”) based on the prescribed dose and one clinically oriented region for evaluation (“whole”) to perform the three-dimensional (3D) global gamma reanalysis. Meanwhile, the percentage gamma passing rate (%GP), mean gamma index (μGI) based on 3%/2 mm criteria, and percentage dose error (%DE) of the dose–volume histogram (DVH) metrics were recorded by COMPASS application. Secondly, the Pearson’s correlation coefficient was used to analyze the correlation between %GP and %DE and between μGI and %DE in different regions. Additionally, receiver operating characteristic (ROC) methodology was applied to quantify the fraction of patient-specific plans evaluated as “fail” and “pass”. In order to improve the correlation between gamma analysis result and clinical criteria, ROC analysis was carried out in accordance with hybridization analysis criteria (%DE ≤3%, %GP ≥90% and %DE ≤3%, μGI ≤0.6). ROC was performed for two purposes: 1) to analyze the sensitivity and specificity of %GP and μGI in different regions for evaluation and 2) to define the ALs of %GP and μGI in the optimal region for evaluation. Finally, the plans introduced with MLC errors were prepared for validation. Moreover, we also compared the positive rate of ALs of both %GP and μGI in detecting MLC error-introduced plans in different regions.Results1) In our study, a number of DVH-based metrics were found to be correlated with the evaluation parameters. The corresponding number was 4, 2, 1, 1, 1, 1, and 3 in γwhole, γ10%, γ20%, γ30%, γ40%, γ50%, and γ60%, respectively, and 5, 3, 0, 1, 1, 4, and 2 in μGIwhole, μGI10%, μGI20%, μGI30%, μGI40%, μGI50%, and μGI60%, respectively. The results by COMPASS have revealed that the %DE of specific structures has a slightly higher correlation with both %GP and μGI of the “whole” region compared with that of any other region. However, it is a moderate correlation (0.5 ≤ |r| < 0.8). 2) The areas under the curves (AUCs) of γwhole, μGIwhole, μGI40%, μGI50%, and μGI60% were >0.7 based on 3%/2 mm criteria. According to the Youden coefficient, we defined the ALs of γwhole ≥92%, μGIwhole ≤0.36, μGI40% ≤0.43, and μGI60% ≤0.40 based on 3%/2 mm criteria. 3) In the validation, for original plans, the accuracy of ALγwhole, ALγ10%, ALμGIwhole, ALμGI40%, ALμGI50%, and ALμGI60% was 23%, 9.8%, 90%, 80.3%, 9.8%, and 88.5%, respectively. For test plans with systematic MLC errors smaller than 0.8 mm, the positive rates of ALγwhole, ALγ10%, ALμGIwhole, ALμGI40%, ALμGI50%, and ALμGI60% were 25%, 58%, 92%, 92%, 42%, and 100%, respectively. For the plans with systematic MLC errors higher than 0.8 mm, the positive rates of all AL%GP and ALμGI were 100%. From the COMPASS validation results, the accuracy of γwhole, μGIwhole, μGI40%, and μGI60% was higher than that of the conventional γ10% and commonly used μGI50%.ConclusionsCompared with the traditional evaluation region (i.e., the criteria with a threshold of 10% or a threshold of 50%, it was the same with the isodose regions of “≥10%”, “≥50%” based on the prescribed dose in our study), the “whole” region is more meaningful to the clinic by COMPASS. The accuracy of μGIwhole is higher than that of the conventional γ10% and the commonly used μGI50%.

Assessment of Statistical Process Control Based DVH Action Levels for Systematic Multi-Leaf Collimator Errors in Cervical Cancer RapidArc Plans.

BackgroundIn the patient-specific quality assurance (QA), DVH is a critical clinically relevant parameter that is finally used to determine the safety and effectiveness of radiotherapy. However, a consensus on DVH-based action levels has not been reached yet. The aim of this study is to explore reasonable DVH-based action levels and optimal DVH metrics in detecting systematic MLC errors for cervical cancer RapidArc plans.MethodsIn this study, a total of 148 cervical cancer RapidArc plans were selected and measured with COMPASS 3D dosimetry system. Firstly, the patient-specific QA results of 110 RapidArc plans were retrospectively reviewed. Then, DVH-based action limits (AL) and tolerance limits (TL) were obtained by statistical process control. Secondly, systematic MLC errors were introduced in 20 RapidArc plans, generating 380 modified plans. Then, the dose difference (%DE) in DVH metrics between modified plans and original plans was extracted from measurement results. After that, the linear regression model was used to investigate the detection limits of DVH-based action levels between %DE and systematic MLC errors. Finally, a total of 180 test plans (including 162 error-introduced plans and 18 original plans) were prepared for validation. The error detection rate of DVH-based action levels was compared in different DVH metrics of 180 test plans.ResultsA linear correlation was found between systematic MLC errors and %DE in all DVH metrics. Based on linear regression model, the systematic MLC errors between -0.94 mm and 0.88 mm could be caught by the TL of PTV95 ([-1.54%, 1.51%]), and the systematic MLC errors between -1.00 mm and 0.80 mm could also be caught by the TL of PTVmean ([-2.06%, 0.38%]). In the validation, for original plans, PTV95 showed the minimum error detection rate of 5.56%. For error-introduced plans with systematic MLC errors more than 1mm, PTVmean showed the maximum error detection rate of 88.89%, and then was followed by PTV95 (86.67%). All the TL of DVH metrics showed a poor error detection rate in identifying error-induced plans with systematic MLC errors less than 1mm.ConclusionIn 3D quality assurance of cervical cancer RapidArc plans, process-based tolerance limits showed greater advantages in distinguishing plans introduced with systematic MLC errors more than 1mm, and reasonable DVH-based action levels can be acquired through statistical process control. During DVH-based verification, main focus should be on the DVH metrics of target volume. OARs in low-dose regions were found to have a relatively higher dose sensitivity to smaller systematic MLC errors, but may be accompanied with higher false error detection rate.

Comparison of pretreatment VMAT quality assurance with the integral quality monitor (IQM) and electronic portal imaging device (EPID).

The purpose of this study was to compare pretreatment volumetric modulated arc therapy (VMAT) quality assurance (QA) measurements and evaluate the multileaf collimator (MLC) error sensitivity of two detectors: the integral quality monitor (IQM) system (iRT systems IQM) and the electronic portal imaging device (EPID) (Varian PortalVision aS1200). Pretreatment QA measurements were performed for 20 retrospective VMAT plans (53 arcs). A subset of ten plans (23 arcs) was used to investigate MLC error sensitivity of each device. Eight MLC error plans were created for each VMAT plan. The errors included systematic opening/closing (±0.25, ±0.50, ±0.75 mm) of the MLC and random positional errors (1 mm) for individual/groups of leaves. The IQM was evaluated using the percent error of the measured cumulative signal relative to the calculated signal. The EPID was evaluated using two methods: a novel percent error of the measured relative to the predicted cumulative signals, and gamma (γ) analysis (1%/1 mm, 2%/2 mm, 3%/3 mm and 3%/1 mm for Stereotactic Body Radiation Therapy plans). The average change in maximum dose obtained from dose‐volume histogram (DVH) data and change in detector signals for different systematic MLC shifts was also compared. Cumulative signal differences showed similar levels of agreement between measured and expected detector signals (IQM: 1.00 ± 0.55%; EPID: 1.22 ± 0.92%). Results from γ analysis lacked specificity. Only the 1%/1 mm criteria produced data with remarkable differences. A strong linear correlation was observed between IQM and EPID cumulative signal differences with MLC error magnitude (R = 0.99). Likewise, results indicate a strong correlation between the cumulative signal for both detectors and DVH dose (RIQM = 0.99; REPID = 0.97). In conclusion, use of cumulative signal differences could be more useful for detecting errors in treatment delivery in EPID than γ analysis.

Effect of Multileaf Collimator Leaf Position Error Determined by Picket Fence Test on Gamma Index Value in Patient-Specific Quality Assurance of Volumetric-Modulated Arc Therapy Plans.

AimThe correlation between the MLC QA (IBA Dosimetry, Germany) results of the picket fence test created with intentional errors and the patient's quality assurance (QA) evaluation was investigated to assess the impact of multileaf collimator (MLC) positioning error on patient QA.Materials and methodsThe picket fence, including error-free and intentional MLC errors, defined in Bank In, Bank Out, and Bank Both were analyzed using MLC QA. The QA of 15 plans consisting of stereotactic radiosurgery (SRS), stereotactic body radiotherapy (SBRT), and conventionally fractionated volumetric-modulated arc therapy (VMAT) acquired with electronic portal imaging devices (EPID) was evaluated in the presence of error-free and MLC errors. The QA of plans were analyzed with 2%/2 mm and 3%/3 mm criteria.ResultsThe passing rates of the picket fence test were 97%, 92%, 91%, and 87% for error-free and intentional errors. The criterion of 3%/3 mm wasn’t able to detect an MLC error for either SRS/SBRT or conventionally fractionated VMAT. The criterion of 2%/2mm was more sensitive to detect MLC error for the conventionally fractionated VMAT than SRS/SBRT. While only two of SBRT plans had <90%, four of conventionally fractionated VMAT plans had a <90% passing rate.ConclusionWe found that the systematic MLC positioning errors defined with picket fence have a smaller but measurable impact on SRS/SBRT than the VMAT plan for a conventionally fractionated and relatively complex plan such as head and neck and endometrium cases.

Deep learning for patient-specific quality assurance: Identifying errors in radiotherapy delivery by radiomic analysis of gamma images with convolutional neural networks.

Patient-specific quality assurance (QA) for intensity-modulated radiation therapy (IMRT) is a ubiquitous clinical procedure, but conventional methods have often been criticized as being insensitive to errors or less effective than other common physics checks. Recently, there has been interest in the application of radiomics, quantitative extraction of image features, to radiotherapy QA. In this work, we investigate a deep learning approach to classify the presence or absence of introduced radiotherapy treatment delivery errors from patient-specific QA. Planar dose maps from 186 IMRT beams from 23 IMRT plans were evaluated. Each plan was transferred to a cylindrical phantom CT geometry. Three sets of planar doses were exported from each plan corresponding to (a) the error-free case, (b) a random multileaf collimator (MLC) error case, and (c) a systematic MLC error case. Each plan was delivered to the electronic portal imaging device (EPID), and planned and measured doses were used to calculate gamma images in an EPID dosimetry software package (for a total of 558 gamma images). Two radiomic approaches were used. In the first, a convolutional neural network with triplet learning was used to extract image features from the gamma images. In the second, a handcrafted approach using texture features was used. The resulting metrics from both approaches were input into four machine learning classifiers (support vector machines, multilayer perceptrons, decision trees, and k-nearest-neighbors) in order to determine whether images contained the introduced errors. Two experiments were considered: the two-class experiment classified images as error-free or containing any MLC error, and the three-class experiment classified images as error-free, containing a random MLC error, or containing a systematic MLC error. Additionally, threshold-based passing criteria were calculated for comparison. In total, 303 gamma images were used for model training and 255 images were used for model testing. The highest classification accuracy was achieved with the deep learning approach, with a maximum accuracy of 77.3% in the two-class experiment and 64.3% in the three-class experiment. The performance of the handcrafted approach with texture features was lower, with a maximum accuracy of 66.3% in the two-class experiment and 53.7% in the three-class experiment. Variability between the results of the four machine learning classifiers was lower for the deep learning approach vs the texture feature approach. Both radiomic approaches were superior to threshold-based passing criteria. Deep learning with convolutional neural networks can be used to classify the presence or absence of introduced radiotherapy treatment delivery errors from patient-specific gamma images. The performance of the deep learning network was superior to a handcrafted approach with texture features, and both radiomic approaches were better than threshold-based passing criteria. The results suggest that radiomic QA is a promising direction for clinical radiotherapy.

Error Detection in Intensity-Modulated Radiation Therapy Quality Assurance Using Radiomic Analysis of Gamma Distributions

To improve the detection of errors in intensity-modulated radiation therapy (IMRT) with a novel method that uses quantitative image features from radiomics to analyze gamma distributions generated during patient specific quality assurance (QA). One hundred eighty-six IMRT beams from 23 patient treatments were delivered to a phantom and measured with electronic portal imaging device dosimetry. The treatments spanned a range of anatomic sites; half were head and neck treatments, and the other half were drawn from treatments for lung and rectal cancers, sarcoma, and glioblastoma. Planar gamma distributions, or gamma images, were calculated for each beam using the measured dose and calculated doses from the 3-dimensional treatment planning system under various scenarios: a plan without errors and plans with either simulated random or systematic multileaf collimator mispositioning errors. The gamma images were randomly divided into 2 sets: a training set for model development and testing set for validation. Radiomic features were calculated for each gamma image. Error detection models were developed by training logistic regression models on these radiomic features. The models were applied to the testing set to quantify their predictive utility, determined by calculating the area under the curve (AUC) of the receiver operator characteristic curve, and were compared with traditional threshold-based gamma analysis. The AUC of the random multileaf collimator mispositioning model on the testing set was 0.761 compared with 0.512 for threshold-based gamma analysis. The AUC for the systematic mispositioning model was 0.717 versus 0.660 for threshold-based gamma analysis. Furthermore, the models could discriminate between the 2 types of errors simulated here, exhibiting AUCs of approximately 0.5 (equivalent to random guessing) when applied to the error they were not designed to detect. The feasibility of error detection in patient-specific IMRT QA using radiomic analysis of QA images has been demonstrated. This methodology represents a substantial step forward for IMRT QA with improved sensitivity and specificity over current QA methods and the potential to distinguish between different types of errors.

Patient-Specific Quality Assurance Using Monte Carlo Dose Calculation and Elekta Log Files for Prostate Volumetric-Modulated Arc Therapy.

Log file–based methods are attracting increasing interest owing to their ability to validate volumetric-modulated arc therapy outputs with high resolution in the leaf and gantry positions and in delivered dose. Cross-validation of these methods for comparison with measurement-based methods using the ionization chamber/ArcCHECK-3DVH software (version 3.2.0) under the same conditions of treatment anatomy and plan enables an efficient evaluation of this method. In this study, with the purpose of cross-validation, we evaluate the accuracy of a log file–based method using Elekta log files and an X-ray voxel Monte Carlo dose calculation technique in the case of leaf misalignment during prostate volumetric-modulated arc therapy. In this study, 10 prostate volumetric-modulated arc therapy plans were used. Systematic multileaf collimator leaf positional errors (±0.4 and ±0.8 mm for each single bank) were deliberately introduced into the optimized plans. Then, the delivered 3-dimensional doses to a phantom with a certain patient anatomy were estimated by our system. These doses were compared with the ionization chamber dose and the ArcCHECK-3DVH dose. For the given phantom and patient anatomy, the estimated dose strongly coincided with the ionization chamber/ArcCHECK-3DVH dose (P < .01). In addition, good agreement between the estimated dose and the ionization chamber/ArcCHECK-3DVH dose was observed. The dose estimation accuracy of our system, which combines Elekta log files and X-ray voxel Monte Carlo dose calculation, was evaluated.

Commissioning and quality assurance for VMAT delivery systems: An efficient time-resolved system using real-time EPID imaging.

An ideal commissioning and quality assurance (QA) program for Volumetric Modulated Arc Therapy (VMAT) delivery systems should assess the performance of each individual dynamic component as a function of gantry angle. Procedures within such a program should also be time-efficient, independent of the delivery system and be sensitive to all types of errors. The purpose of this work is to develop a system for automated time-resolved commissioning and QA of VMAT control systems which meets these criteria. The procedures developed within this work rely solely on images obtained, using an electronic portal imaging device (EPID) without the presence of a phantom. During the delivery of specially designed VMAT test plans, EPID frames were acquired at 9.5Hz, using a frame grabber. The set of test plans was developed to individually assess the performance of the dose delivery and multileaf collimator (MLC) control systems under varying levels of delivery complexities. An in-house software tool was developed to automatically extract features from the EPID images and evaluate the following characteristics as a function of gantry angle: dose delivery accuracy, dose rate constancy, beam profile constancy, gantry speed constancy, dynamic MLC positioning accuracy, MLC speed and acceleration constancy, and synchronization between gantry angle, MLC positioning and dose rate. Machine log files were also acquired during each delivery and subsequently compared to information extracted from EPID image frames. The largest difference between measured and planned dose at any gantry angle was 0.8% which correlated with rapid changes in dose rate and gantry speed. For all other test plans, the dose delivered was within 0.25% of the planned dose for all gantry angles. Profile constancy was not found to vary with gantry angle for tests where gantry speed and dose rate were constant, however, for tests with varying dose rate and gantry speed, segments with lower dose rate and higher gantry speed exhibited less profile stability. MLC positional accuracy was not observed to be dependent on the degree of interdigitation. MLC speed was measured for each individual leaf and slower leaf speeds were shown to be compensated for by lower dose rates. The test procedures were found to be sensitive to 1mm systematic MLC errors, 1mm random MLC errors, 0.4mm MLC gap errors and synchronization errors between the MLC, dose rate and gantry angle controls systems of 1°. In general, parameters measured by both EPID and log files agreed with the plan, however, a greater average departure from the plan was evidenced by the EPID measurements. QA test plans and analysis methods have been developed to assess the performance of each dynamic component of VMAT deliveries individually and as a function of gantry angle. This methodology relies solely on time-resolved EPID imaging without the presence of a phantom and has been shown to be sensitive to a range of delivery errors. The procedures developed in this work are both comprehensive and time-efficient and can be used for streamlined commissioning and QA of VMAT delivery systems.

The dosimetric impact of control point spacing for sliding gap MLC fields.

Dynamic sliding gap multileaf collimator (MLC) fields are used to model MLC properties within the treatment planning system (TPS) for dynamic treatments. One of the key MLC properties in the Eclipse TPS is the dosimetric leaf gap (DLG) and precise determination of this parameter is paramount to ensuring accurate dose delivery. In this investigation, we report on how the spacing between control points (CPs) for sliding gap fields impacts the dose delivery, MLC positioning accuracy, and measurement of the DLG. The central axis dose was measured for sliding gap MLC fields with gap widths ranging from 2 to 40 mm. It was found that for deliveries containing two CPs, the central axis dose was underestimated by the TPS for all gap widths, with the maximum difference being 8% for a 2 mm gap field. For the same sliding gap fields containing 50 CPs, the measured dose was always within ±2% of the TPS dose. By directly measuring the MLC trajectories we show that this dose difference is due to a systematic MLC gap error for fields containing two CPs, and that the cause of this error is due to the leaf position offset table which is incorrectly applied when the spacing between CPs is too large. This MLC gap error resulted in an increase in the measured DLG of 0.5 mm for both 6 MV and 10 MV, when using fields with 2 CPs compared to 50 CPs. Furthermore, this change in DLG was shown to decrease the mean TPS‐calculated dose to the target volume by 2.6% for a clinical IMRT test plan. This work has shown that systematic MLC positioning errors occur for sliding gap MLC fields containing two CPs and that using these fields to model critical TPS parameters, such as the DLG, may result in clinically significant systematic dose calculation errors during subsequent dynamic MLC treatments.PACS number(s): 87.56.nk

An EPID‐based system for gantry‐resolved MLC quality assurance for VMAT

Multileaf collimator (MLC) positions should be precisely and independently measured as a function of gantry angle as part of a comprehensive quality assurance (QA) program for volumetric‐modulated arc therapy (VMAT). It is also ideal that such a QA program has the ability to relate MLC positional accuracy to patient‐specific dosimetry in order to determine the clinical significance of any detected MLC errors. In this work we propose a method to verify individual MLC trajectories during VMAT deliveries for use as a routine linear accelerator QA tool. We also extend this method to reconstruct the 3D patient dose in the treatment planning system based on the measured MLC trajectories and the original DICOM plan file. The method relies on extracting MLC positions from EPID images acquired at 8.41 fps during clinical VMAT deliveries. A gantry angle is automatically tagged to each image in order to obtain the MLC trajectories as a function of gantry angle. This analysis was performed for six clinical VMAT plans acquired at monthly intervals for three months. The measured trajectories for each delivery were compared to the MLC positions from the DICOM plan file. The maximum mean error detected was 0.07 mm and a maximum root‐mean‐square error was 0.8 mm for any leaf of any delivery. The sensitivity of this system was characterized by introducing random and systematic MLC errors into the test plans. It was demonstrated that the system is capable of detecting random and systematic errors on the range of 1–2 mm and single leaf calibration errors of 0.5 mm. The methodology developed in the work has potential to be used for efficient routine linear accelerator MLC QA and pretreatment patient‐specific QA and has the ability to relate measured MLC positional errors to 3D dosimetric errors within a patient volume.PACS number(s): 87.55.Qr

TH-AB-202-02: Real-Time Verification and Error Detection for MLC Tracking Deliveries Using An Electronic Portal Imaging Device

Purpose: The added complexity of the real-time adaptive multi-leaf collimator (MLC) tracking increases the likelihood of undetected MLC delivery errors. In this work we develop and test a system for real-time delivery verification and error detection for MLC tracking radiotherapy using an electronic portal imaging device (EPID). Methods: The delivery verification system relies on acquisition and real-time analysis of transit EPID image frames acquired at 8.41 fps. In-house software was developed to extract the MLC positions from each image frame. Three comparison metrics were used to verify the MLC positions in real-time: (1) field size, (2) field location and, (3) field shape. The delivery verification system was tested for 8 VMAT MLC tracking deliveries (4 prostate and 4 lung) where real patient target motion was reproduced using a Hexamotion motion stage and a Calypso system. Sensitivity and detection delay was quantified for various types of MLC and system errors. Results: For both the prostate and lung test deliveries the MLC-defined field size was measured with an accuracy of 1.25 cm2 (1 SD). The field location was measured with an accuracy of 0.6 mm and 0.8 mm (1 SD) for lung and prostate respectively. Field location errors (i.e. tracking in wrong direction) with a magnitude of 3 mm were detected within 0.4 s of occurrence in the X direction and 0.8 s in the Y direction. Systematic MLC gap errors were detected as small as 3 mm. The method was not found to be sensitive to random MLC errors and individual MLC calibration errors up to 5 mm. Conclusion: EPID imaging may be used for independent real-time verification of MLC trajectories during MLC tracking deliveries. Thresholds have been determined for error detection and the system has been shown to be sensitive to a range of delivery errors.

Quality assurance for online adapted treatment plans: benchmarking and delivery monitoring simulation.

An important challenge facing online adaptive radiation therapy is the development of feasible and efficient quality assurance (QA). This project aimed to validate the deliverability of online adapted plans and develop a proof-of-concept online delivery monitoring system for online adaptive radiation therapy QA. The first part of this project benchmarked automatically online adapted prostate treatment plans using traditional portal dosimetry IMRT QA. The portal dosimetry QA results of online adapted plans were compared to original (unadapted) plans as well as randomly selected prostate IMRT plans from our clinic. In the second part, an online delivery monitoring system was designed and validated via a simulated treatment with intentional multileaf collimator (MLC) errors. This system was based on inputs from the dynamic machine information (DMI), which continuously reports actual MLC positions and machine monitor units (MUs) at intervals of 50 ms or less during delivery. Based on the DMI, the system performed two levels of monitoring/verification during the delivery: (1) dynamic monitoring of cumulative fluence errors resulting from leaf position deviations and visualization using fluence error maps (FEMs); and (2) verification of MLC positions against the treatment plan for potential errors in MLC motion and data transfer at each control point. Validation of the online delivery monitoring system was performed by introducing intentional systematic MLC errors (ranging from 0.5 to 2 mm) to the DMI files for both leaf banks. These DMI files were analyzed by the proposed system to evaluate the system's performance in quantifying errors and revealing the source of errors, as well as to understand patterns in the FEMs. In addition, FEMs from 210 actual prostate IMRT beams were analyzed using the proposed system to further validate its ability to catch and identify errors, as well as establish error magnitude baselines for prostate IMRT delivery. Online adapted plans were found to have similar delivery accuracy in comparison to clinical IMRT plans when validated with portal dosimetry IMRT QA. FEMs for the simulated deliveries with intentional MLC errors exhibited distinct patterns for different MLC error magnitudes and directions, indicating that the proposed delivery monitoring system is highly specific in detecting the source of errors. Implementing the proposed QA system for online adapted plans revealed excellent delivery accuracy: over 99% of leaf position differences were within 0.5 mm, and >99% of pixels in the FEMs had fluence errors within 0.5 MU. Patterns present in the FEMs and MLC control point analysis for actual patient cases agreed with the error pattern analysis results, further validating the system's ability to reveal and differentiate MLC deviations. Calculation of the fluence map based on the DMI was performed within 2 ms after receiving each DMI input. The proposed online delivery monitoring system requires minimal additional resources and time commitment to the current clinical workflow while still maintaining high sensitivity to leaf position errors and specificity to error types. The presented online delivery monitoring system therefore represents a promising QA system candidate for online adaptive radiation therapy.

Sensitivity of volumetric modulated arc therapy patient specific QA results to multileaf collimator errors and correlation to dose volume histogram based metrics

This study investigates the impact of systematic multileaf collimator (MLC) positional errors on gamma analysis results used for quality assurance (QA) of Rapidarc treatments. In addition, this study evaluates the relationship of these gamma analysis results and clinical dose volume histogram metrics (DVH) for Rapidarc treatment plans. Five prostate plans were modified by the introduction of systematic MLC errors. The MLC shifts to each individual active leaf introduced were 0.25, 0.5, 0.75, and 1 mm. All QA verification plans were delivered and estimated 3D patient dose or high density phantom dose were obtained based on the ArcCHECK measurement files. QA gamma analysis of 3%/3 mm and 2%/2 mm were implemented and relationships to dose differences in DVH metrics encountered due to MLC errors were determined. Tolerances of 3% and 5% for DVH metric were implemented to determine the sensitivity of gamma analysis to MLC errors. A calculation of sensitivity was determined from the number of incidences of false negative and false positive cases in gamma analysis results. The sensitivity of global gamma analysis for criteria of 3%/3 mm was 0.78 and for 2%/2 mm was 0.82. A number of instances occurred for an acceptable VMAT QA gamma index which did not indicate a DVH metric dose error greater than 5%. The correlation between global gamma analysis using criteria 3%/3 mm and DVH metric dose error were all <0.8 indicating less than a strong correlation. There is a greater sensitivity for detection of dosimetric errors occurring in a Rapidarc plan using gamma criteria of 2%/2 mm than 3%/3 mm. However, there is lack of consistently strong correlation between global gamma indexes and clinical DVH metrics for PTV and bladder and rectum for Rapidarc plans. It is recommended that the sole use of gamma index for Rapidarc QA plan evaluation could be insufficient and a methodology for evaluation of delivered dose to patient is required.

Effect of MLC leaf position, collimator rotation angle, and gantry rotation angle errors on intensity-modulated radiotherapy plans for nasopharyngeal carcinoma

The purpose of this study was to investigate the effect of multileaf collimator (MLC) leaf position, collimator rotation angle, and accelerator gantry rotation angle errors on intensity-modulated radiotherapy plans for nasopharyngeal carcinoma. To compare dosimetric differences between the simulating plans and the clinical plans with evaluation parameters, 6 patients with nasopharyngeal carcinoma were selected for simulation of systematic and random MLC leaf position errors, collimator rotation angle errors, and accelerator gantry rotation angle errors. There was a high sensitivity to dose distribution for systematic MLC leaf position errors in response to field size. When the systematic MLC position errors were 0.5, 1, and 2mm, respectively, the maximum values of the mean dose deviation, observed in parotid glands, were 4.63%, 8.69%, and 18.32%, respectively. The dosimetric effect was comparatively small for systematic MLC shift errors. For random MLC errors up to 2mm and collimator and gantry rotation angle errors up to 0.5°, the dosimetric effect was negligible. We suggest that quality control be regularly conducted for MLC leaves, so as to ensure that systematic MLC leaf position errors are within 0.5mm. Because the dosimetric effect of 0.5° collimator and gantry rotation angle errors is negligible, it can be concluded that setting a proper threshold for allowed errors of collimator and gantry rotation angle may increase treatment efficacy and reduce treatment time.

Biological consequences of MLC calibration errors in IMRT delivery and QA

The purpose of this work is threefold: (1) to explore biological consequences of the multileaf collimator (MLC) calibration errors in intensity modulated radiotherapy (IMRT) of prostate and head and neck cancers, (2) to determine levels of planning target volume (PTV) and normal tissue under- or overdose flagged with clinically used QA action limits, and (3) to provide biologically based input for MLC QA and IMRT QA action limits. Ten consecutive prostate IMRT cases and ten consecutive head and neck IMRT cases were used. Systematic MLC offsets (i.e., calibration error) were introduced for each control point of the plan separately for X1 and X2 leaf banks. Offsets were from - 2 to 2 mm with a 0.5 mm increment. The modified files were imported into the planning system for forward dose recalculation. The original plan served as the reference. The generalized equivalent uniform dose (gEUD) was used as the biological index for the targets, rectum, parotid glands, brainstem, and spinal cord. Each plan was recalculated on a CT scan of a 27 cm diameter cylindrical phantom with a contoured 0.6 cc ion chamber. Dose to ion chamber and 3D gamma analysis were compared to the reference plan. QA pass criteria: (1) at least 95% of voxels with a dose cutoff of 50% of maximum dose have to pass at 3 mm/3% and (2) dose to chamber within 2% of the reference dose. For prostate cases, differences in PTV and rectum gEUD greater than 2% were identified. However, a larger proportion of plans leading to greater than 2% difference in prostate PTV gEUD passed the ion chamber QA but not 3D gamma QA. A similar trend was found for the rectum gEUD. For head and neck IMRT, the QA pass criteria flagged plans leading to greater than 4% differences in PTV gEUD and greater than 5% differences in the maximum dose to brainstem. If pass criteria were relaxed to 90% for gamma and 3% for ion chamber QA, plans leading to a 5% difference in PTV gEUD and a 5%-8% difference in brainstem maximum dose would likely pass IMRT QA. A larger proportion of head and neck plans with greater than 2% PTV gEUD difference passed 3D gamma QA compared to ion chamber QA. For low modulation plans, there is a better chance to catch MLC calibration errors with 3D gamma QA rather than ion chamber QA. Conversely, for high modulation plans, there is a better chance to catch MLC calibration errors with ion chamber QA rather than with 3D gamma QA. Ion chamber and 3D gamma analysis IMRT QA can detect greater than 2% change in gEUD for PTVs and critical structures for low modulation treatment plans. For high modulation treatment plans, ion chamber and 3D gamma analysis can detect greater than 2% change in gEUD for PTVs and a 5% change in critical structure gEUD since either QA methods passes the QA criteria. For gEUD changes less than those listed above, either QA method has the same proportion of passing rate.

Understanding the impact of RapidArc therapy delivery errors for prostate cancer

The purpose of this study is to simulate random and systematic RapidArc delivery errors for external beam prostate radiotherapy plans in order to determine the dose sensitivity for each error type. Ten prostate plans were created with a single 360° arc. The DICOM files for these treatment plans were then imported into an in‐house computer program that introduced delivery errors. Random and systematic gantry position (0.25°, 0.5°, 1°), monitor unit (MU) (1.25%, 2.5%, 5%), and multileaf collimator (MLC) position (0.5, 1, 2 mm) errors were introduced. The MLC errors were either random or one of three types of systematic errors, where the MLC banks moved in the same (MLC gaps remain unchanged) or opposing directions (increasing or decreasing the MLC gaps). The generalized equivalent uniform dose (gEUD) was calculated for the original plan and all treatment plans with errors introduced. The dose sensitivity for the cohort was calculated using linear regression for the gantry position, MU, and MLC position errors. Because there was a large amount of variability for systematic MLC position errors, the dose sensitivity of each plan was calculated and correlated with plan MU, mean MLC gap, and the percentage of MLC leaf gaps less than 1 and 2 cm for each individual plan. We found that random and systematic gantry position errors were relatively insignificant (< 0.1% gEUD change) for gantry errors up to 1°. Random MU errors were also insignificant, and systematic MU increases caused a systematic increase in gEUD. For MLC position errors, random MLC errors were relatively insignificant up to 2 mm as had been determined in previous IMRT studies. Systematic MLC shift errors caused a decrease of approximately −1% in the gEUD per mm. For systematic MLC gap open errors, the dose sensitivity was 8.2%/mm and for MLC gap close errors the dose sensitivity was −7.2%/mm. There was a large variability for MLC gap open/close errors for the ten RapidArc plans which correlated strongly with MU, mean gap width, and percentage of MLC gaps less than 1 or 2 cm. This study evaluates the magnitude of various simulated RapidArc delivery errors by calculating gEUED on various prostate plans.PACS numbers: 87.55.x, 87.55.D, 87.55.de, 87.55.dk

Clinical significance of multi-leaf collimator positional errors for volumetric modulated arc therapy

Background and purpose Multi-leaf collimator (MLC) positional errors occur during intensity modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT) deliveries. The impact of such errors has been evaluated for IMRT but not VMAT. The purpose of this work is to understand how random and systematic VMAT MLC positional errors affect the patient dose distribution. Materials and methods Eight head and neck single arc (360°) VMAT treatment plans were created. Random and two types of systematic MLC errors were simulated for error magnitudes of 0.25, 0.5, 1, 2 and 5 mm. The two types of systematic MLC errors were: (1) MLC banks are shifted in the same direction (left or right) and (2) MLC banks are shifted in opposing directions resulting in smaller or larger field shapes. The MLC errors were simulated, for all control points, on both banks of active MLC leaves only. Results There is a linear correlation of MLC errors with gEUD for all error types. The gEUD dose sensitivities with MLC error for the PTV70 were −0.2, −0.9, −2.8 and 1.9 Gy/mm for random, systematic shift, systematic close and systematic open MLC errors, respectively. The sensitivity of VMAT plans to MLC positional errors was similar to those of IMRT plans with less than 50 segments but much less than those created for a step and shoot with more than 50 segments or sliding-window delivery technique. To maintain the PTV70 to within 2% would require that MLC open/close errors be within 0.6 mm. Conclusions Radiation therapy centers should have adequate quality assurance programs in place to assess open/close MLC errors (i.e. leaf gap errors) as they tend to be more impactful than random or systematic MLC shift errors.

Impact of MLC leaf position errors on simple and complex IMRT plans for head and neck cancer**Presented at the 48th Annual Meeting of the American Association of Physicists in Medicine, 30 July 2006.

The dosimetric impact of random and systematic multi-leaf collimator (MLC) leaf position errors is relatively unknown for head and neck intensity-modulated radiotherapy (IMRT) patients. In this report we studied 17 head and neck IMRT patients, including 12 treated with simple plans (<50 segments) and 5 treated with complex plans (>100 segments). Random errors (−2 to +2 mm) and systematic errors (±0.5 mm and ±1 mm) in MLC leaf positions were introduced into the clinical plans and the resultant dose distributions were analyzed based on defined endpoint doses. The dosimetric effect was insignificant for random MLC leaf position errors up to 2 mm for both simple and complex plans. However, for systematic MLC leaf position errors, we found significant dosimetric differences between the simple and complex IMRT plans. For 1 mm systematic error, the average changes in D95% were 4% in simple plans versus 8% in complex plans. The average changes in D0.1 cc of the spinal cord and brain stem were 4% in simple plans versus 12% in complex plans. The average changes in parotid glands were 9% in simple plans versus 13% for the complex plans. Overall, simple IMRT plans are less sensitive to leaf position errors than complex IMRT plans.