Prostate Volumetric Modulated Arc Therapy Research Articles

Optimizing volumetric modulated arc therapy prostate planning using an automated Fine-Tuning process through dynamic adjustment of optimization parameters

In radiotherapy treatment planning, optimization is essential for achieving the most favorable plan by adjusting optimization criteria. This study introduced an innovative approach to automatically fine-tune optimization parameters for volumetric modulated arc therapy prostate planning, ensuring all constraints were met. A knowledge-based planning model was invoked, and the fine-tuning process was applied through an in-house developed script. Among 25 prostate plans, this fine-tuning increased the number of plans meeting all constraints from 10/25 to 22/25, with a reduction in mean monitor units per gray without increasing plan’s complexity. This automation improved efficiency by saving time and resources in treatment planning.

Evaluation of the trend of set-up errors during the treatment period using set-up margin in prostate radiotherapy

Accurate information on set-up error during radiotherapy is essential for determining the optimal number of treatments in hypofractionated radiotherapy for prostate cancer. This necessitates careful control by the radiotherapy staff to assess the patient's condition. This study aimed to develop an evaluation method of the temporal trends in a patient's specific prostate movement during treatment using image matching and margin values. This study included 65 patients who underwent prostate volumetric modulated arc therapy (mean treatment time, 87.2 s). Set-up errors were assessed using bone, inter-, and intra-fraction marker matching across 39 fractions. The set-up margin was determined by dividing the four periods into 39 fractions using Stroom's formula and correlation coefficient. The intra-fraction set-up error was biased in the anterior-superior (AS) direction during treatment. The temporal trend of set-up errors during radiotherapy slightly increased based on bone matching and inter-fraction marker matching, with a 1.6-mm difference in the set-up margin fractions 11 to 20. The correlation coefficient of the mean prostate movement during treatment significantly decreased in the superior-inferior direction, while remaining high in the left-right and anterior-posterior directions. Image matching contributed significantly to the improvement of set-up errors; however, careful attention is needed for prostate movement in the AS direction, particularly during short treatment times. Understanding the trend of set-up errors during the treatment period is essential in numerical information sharing on patient condition and evaluating the margins for tailored hypo-fractionated radiotherapy, considering the facility's image-guided radiation therapy technology.

Deep learning-based tools to distinguish plan-specific from generic deviations in EPID-based in vivo dosimetry.

Dose distributions calculated with electronic portal imaging device (EPID)-based in vivo dosimetry (EIVD) differ from planned dose distributions due to generic and plan-specific deviations. Generic deviations are characteristic to a class of plans. Examples include limitations in EIVD dose reconstruction, inaccuracies in treatment planning system (TPS) calculations and systematic machine deviations. Plan-specific deviations have an unpredictable character. Examples include discrepancies between the patient model used for dose calculation and the patient position or anatomy during delivery, random machine deviations, and data transfer, human or software errors. During the inspection work performed with traditional γ-evaluation statistical methods: (i) generic deviations raise alerts that need to be inspected but that rarely lead to action as their root cause is usually understood and (ii) the detection of relevant plan-specific deviations may be hindered by the presence of generic deviations. To investigate whether deep learning-based tools can help in identifying γ-alerts raised by generic deviations and in improving the detectability of plan-specific deviations. A 3D U-Net was trained as an autoencoder to reconstruct underlying patterns of generic deviations in γ-distributions. The network was trained for four treatment disease sites differently affected by generic deviations: volumetric modulated arc therapy (VMAT) lung (no known deviations), VMAT prostate (TPS inaccuracies), VMAT head-and-neck (EIVD limitations) and intensity modulated radiation therapy (IMRT) breast (large EIVD limitations). The network was trained with virtual non-transit γ-distributions: 60 train/10 validation for the VMAT sites and 30 train/10 validation for IMRT breast. It was hypothesized that in vivo γ-distributions obtained in the presence of plan-specific deviations would differ from those seen during training. For each disease site, the sensitivity of γ-analysis and the network to detect (synthetically introduced) patient-related deviations was compared by receiver operator characteristic analysis. The investigated deviations were patient positioning errors, weight gain or loss, and tumor volume changes. The clinical relevance was illustrated qualitatively with 793 in vivo clinical cases (141 lung, 136 head-and-neck, 209 prostate and 307 breast). Error detectability of patient-related deviations was better with the network than with γ-analysis. The average area under the curve values over all sites were 0.86±0.12(1SD) and 0.69±0.25(1SD), respectively. Regarding in vivo clinical results, the percentage of cases differently classified by γ-analysis and the network was 1%, 19%, 18% and 64% for lung, head-and-neck, prostate, and breast, respectively. In head-and-neck and breast cases, 45 γ-only alerts were examined, of which 43 were attributed to EPID dose reconstruction limitations. For prostate, all 15 investigated γ-only alerts were due to known TPS inaccuracies. All 59 investigated network alerts were explained by either patient-related deviations or EPID acquisition incidents. Some patient-related deviations detected by the network were not detected by γ-analysis. Deep learning-based tools trained to reconstruct underlying patterns of generic deviations in γ-distributions can be used to (i) automatically identify false positives within the set of γ-alerts and (ii) improve the detection of plan-specific deviations, hence minimizing the likelihood of false negatives. The presented method provides clear additional value to the γ-alert management process for large scale EIVD systems.

Complexity analysis of VMAT prostate plans: insights from dimensionality reduction and information theory techniques

Abstract Introduction: Volumetric Modulated Arc Therapy (VMAT) is a state-of-the-art prostate cancer treatment, defined by high dose gradients around targets. Its unique dose shaping incurs hidden complexity, impacting treatment deliverability, carcinogenesis, and machine strain. This study compares various aperture-based VMAT complexity indices in prostate cases using principal component and mutual information analyses. It suggests essential properties for an ideal complexity index from an information-theoretic viewpoint. Material and methods: The following ten complexity indices were calculated in 217 VMAT prostate plans: circumference over area (CoA), edge metric (EM), equivalent square field (ESF), leaf travel (LT), leaf travel modulation complexity score for VMAT (LTMCSV), mean-field area (MFA), modulation complexity score (standard MCS and VMAT variant MCSV), plan irregularity (PI), and small aperture score (SAS5mm). Principal component analysis (PCA) was applied to explore the correlations between the metrics. The differential entropy of all metrics was also calculated, along with the mutual information for all 45 metric pairs. Results: Whole-pelvis plans had greater complexity across all indices. The first three principal components explained 96.2% of the total variance. The complexity metrics formed three groups with similar conceptual characteristics, particularly ESF, LT, MFA, PI, and EM, SAS5mm. The differential entropy varied across the complexity metrics (PI having the smallest vs. EM the largest). Mutual information analysis (MIA) confirmed some metrics’ interdependence, although other pairs, such as LTMCSV/SAS5mm, LT/MCSV, and EM/SAS5mm, were found to share minimal MI. Conclusions: There are many complexity indices for VMAT described in the literature. PCA and MIA analyses can uncover significant overlap among them. However, this is not entirely reducible through dimensionality reduction techniques, suggesting that there also exists some reciprocity. When designing predictive models of quality assurance metrics, PCA and MIA may prove useful for feature engineering.

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.

An ultra-fast deep-learning-based dose engine for prostate VMAT via knowledge distillation framework with limited patient data

Objective. Deep-learning (DL)-based dose engines have been developed to alleviate the intrinsic compromise between the calculation accuracy and efficiency of the traditional dose calculation algorithms. However, current DL-based engines typically possess high computational complexity and require powerful computing devices. Therefore, to mitigate their computational burdens and broaden their applicability to a clinical setting where resource-limited devices are available, we proposed a compact dose engine via knowledge distillation (KD) framework that offers an ultra-fast calculation speed with high accuracy for prostate Volumetric Modulated Arc Therapy (VMAT). Approach. The KD framework contains two sub-models: a large pre-trained teacher and a small to-be-trained student. The student receives knowledge transferred from the teacher for better generalization. The trained student serves as the final engine for dose calculation. The model input is patient computed tomography and VMAT dose in water, and the output is DL-calculated patient dose. The ground-truth \\dose was computed by the Monte Carlo module of the Monaco treatment planning system. Twenty and ten prostate cases were included for model training and assessment, respectively. The model’s performance (teacher/student/student-only) was evaluated by Gamma analysis and inference efficiency. Main results. The dosimetric comparisons (input/DL-calculated/ground-truth doses) suggest that the proposed engine can effectively convert low-accuracy doses in water to high-accuracy patient doses. The Gamma passing rate (2%/2 mm, 10% threshold) between the DL-calculated and ground-truth doses was 98.64 ± 0.62% (teacher), 98.13 ± 0.76% (student), and 96.95 ± 1.02% (student-only). The inference time was 16 milliseconds (teacher) and 11 milliseconds (student/student-only) using a graphics processing unit device, while it was 936 milliseconds (teacher) and 374 milliseconds (student/student-only) using a central processing unit device. Significance. With the KD framework, a compact dose engine can achieve comparable accuracy to that of a larger one. Its compact size reduces the computational burdens and computing device requirements, and thus such an engine can be more clinically applicable.

A method for patient-specific DVH verification using a high-sampling-rate log file in an Elekta linac.

We have proposed a method for patient-specific dose-volume histogram (DVH) verification using a 40-ms high-sampling-rate log file (HLF) available in an Elekta linac. Ten prostate volumetric-modulated arc therapy plans were randomly selected, and systematic leaf position errors of ±0.2, ±0.4, or ±0.8mm were added to the 10 plans, thereby producing a total of 70 plans. An RTP file was created by interpolating each leaf position in the HLF to obtain values at each control point, which is subsequently exported to a treatment planning system. The isocenter dose calculated by the HLF-based plan to a phantom (Delta4 Phantom+) was compared to that measured by the diode in the phantom in order to evaluate the accuracy of the HLF-based dose calculation. The D95 of the planning target volume (PTV) was also compared between the HLF-based plans and the original plans with the systematic leaf position errors, the latter being referred to as theory-based plans. Sensitivities of the DVH parameters in the target, the rectum, and the bladder were also calculated with the varied systematic leaf position errors. The relative differences in the isocenter doses between the HLF-based calculations and the measurements among the 70 plans were 0.21%±0.67% (SD). The maximum relative differences in PTV D95 between the HLF-based and the theory-based plans among the 70 cases were 0.11%. The patient-specific DVH verification method detected a change in the target DVH parameters of less than 1% when the systematic leaf position error was ±0.2mm. It is therefore suggested that the proposed DVH verification method may simplify patient-specific dose quality assurance procedures without compromising accuracy and sensitivity.

Knowledge-Based Treatment Planning Model Improves Organ-at-Risk Sparing and Plan Consistency across Clinics for Prostate Radiation Therapy

<h3>Purpose/Objective(s)</h3> The quality of volumetric modulated arc therapy (VMAT) plans highly depends on the planner's preference and experience. A commercially available knowledge-based planning (KBP) module has been widely implemented to estimate the patient-specific optimal dose-volume histogram and enable semi-automatic treatment planning. The purpose of this study was to investigate whether the KBP clinical implementations have improved organ-at-risk (OAR) sparing and plan quality consistency across clinics within our institution for prostate cancer VMAT treatments. <h3>Materials/Methods</h3> A cohort of 1,484 prostate cancer VMAT treatment plans delivered across three different clinics within our institution from January 2016 to January 2022 were retrospectively identified. The prescription (Rx) schemes of 45 Gy and 70 Gy were selected for statistical investigations since these were the only Rx schemes containing more than 50 VMAT plans created with KBP models. Dosimetric parameters of the bladder, rectum, left/right femoral heads, penile bulb, and volume ratio of 50% isodose line to PTV (R50) were extracted. <h3>Results</h3> Since its first adoption in the last quarter of 2018, KBP planning remained at a low application rate from 0% to 20.0% in each quarter. After publishing KBP models clinically at the beginning of 2021, the application rate of prostate KBP models raised to levels from 53.5% to 71.0% in each quarter. For the 70 Gy Rx scheme, parameters of R50 (w/ KBP: 3.67±4.35 vs w/o KBP: 4.17±6.25, <i>p</i>=0.009), bladder (V_40Gy: 25.49%±13.02% vs 30.50%±16.52%, <i>p</i>=0.002; D_mean: 26.47±9.17 Gy vs 30.51±10.11 Gy, <i>p</i>=0.003) and rectum (V_45Gy: 17.45%±9.25% vs 25.23%±10.40%, <i>p</i><0.001; D_mean: 25.41±5.97 Gy vs 30.59±6.77 Gy, <i>p</i><0.001) were all statistically significantly improved with KBP planning. For the 45 Gy Rx scheme, the results were consistent with the 70 Gy Rx scheme. Across the three different clinics, the investigated parameters became more consistent after the adoption of KBP planning (e.g., bladder D_mean in Gy, w/ KBP: 27.52±10.21, 26.47±9.15 and 31.10±9.83 for clinic 1, 2 and 3, respectively, <i>p</i>₁₂=0.646, <i>p</i>₁₃=0.276, <i>p</i>₂₃=0.453; w/o KBP: 28.03±12.52, 27.02±13.97 and 39.85±23.27 for clinic 1, 2 and 3, respectively, <i>p</i>₁₂=0.425, <i>p</i>₁₃<0.001, <i>p</i>₂₃<0.001). This indicates that, with KBP implementations, we have achieved improved OAR-sparing across the institution as well as increased levels of plan quality standardization. <h3>Conclusion</h3> This study demonstrated that KBP models could improve OAR sparing and plan quality consistency across clinics for prostate VMAT radiation therapy. Clinical adoptions of KBP could potentially reduce inter-planner variations and facilitate the standardization of VMAT treatment planning.

Validation of RapidPlan Knowledge-Based Model for Volumetric-Modulated Arc Therapy in Prostate Cancer.

This study aims to investigate the viability of using RapidPlan (RP) knowledge-based (KB) treatment plans to initiate the new prostate volumetric-modulated arc therapy (VMAT) plans. The planning data for 120 prostate VMAT treatment plans were entered into the RP system's database. The database of previous VMAT plans was divided into four model groups for training in the RP system. The models were based on the numbers of 20, 60, and 120 prostate VMAT plans. The model of 120 plans used automated priority and manual priority for the optimization process. The models of 20 and 60 plans used only manual priority for optimization. Each model was validated on 15 cases of new prostate cancer patients by comparing RP model plans against manual clinical plans optimized according to the clinical dose constraints. The RP models can estimate the dose comparable target volume to the manual clinical plan, which evaluated values of Dmax, D95%, D98%, HI, and CI and showed comparable results. For the normal organ doses of the bladder, rectum, penile bulb, and femoral head, all RP models exhibited a comparable or better dose than the manual clinical plan, except for the RP models using the automated priority for the optimization process, which cannot control the rectum dose below the dose constraints. The Varian RP KB planning can produce comparable doses or better doses with the clinical manual in a single optimization, although the RP model uses a minimum requirement of the planning number for the model training. The RP models can enhance the efficacy and quality of plans, which depend on the number of VMAT plans used in RP model training for prostate cancer.

Volumetric dose extension for isodose tuning.

To develop a method that can extend dose from two isodose surfaces (isosurfaces) to the entire patient volume, and to demonstrate its application in radiotherapy plan isodose tuning. We hypothesized that volumetric dose distribution can be extended from two isosurfaces-the 100% isosurface and a reference isosurface-with the distances to these two surfaces ( and ) as extension variables. The extension function is modeled by a three-dimensional lookup table (LUT), where voxel dose values from clinical plans are binned by three indexes: , , and (reference dose level). The mean and standard deviation of voxel doses in each bin are calculated and stored in LUT. Volumetric dose extension is performed voxel-wisely by indexing the LUT with the , , and of each query voxel. The mean dose stored in the corresponding bin is filled into the query voxel as extended dose, and the standard deviation be filled voxel-wisely as the uncertainty of extension result. We applied dose extension in isodose tuning, which aims to tune volumetric dose distribution by isosurface dragging. We adopted extended dose as an approximate dose estimation, and combined it with dose correction strategy to achieve accurate dose tuning. We collected 32 post-operative prostate volumetric-modulated arc therapy (VMAT) cases and built the LUT and its associated uncertainties from the doses of 27 cases. The dose extension method was tested on five cases, whose dose distributions were defined as ground truth (GT). We extended the doses from 100% and 50% GT isosurfaces to the entire volume, and evaluated the accuracy of extended doses. The 5mm/5% gamma passing rate (GPR) of extended doses are 92.0%. The mean error is 4.5%, which is consistent to the uncertainty estimated by LUT. The dose difference in 90.5% of voxels is within two sigma and 97.5% in three sigma. The calculation time is less than 2seconds. To simulate plan isodose tuning, we optimized a dose with less sparing on rectum (than GT dose) and defined it as a "base dose"-the dose awaiting isosurface dragging. In front-end, the simulated isodose tuning is conducted as such that the base dose was given to plan tuner, and its 50% isosurface would be dragged to the desired position (position of 50% isosurface in GT dose). In back-end, the output of isodose tuning is obtained by (1) extending dose from the desired isosurfaces and viewed the extended dose as an approximate dose, (2) obtaining a correction map from the base dose, and (3) applying the correction map to the extended dose. The accuracy of output-extended dose with correction-was 97.2% in GPR (3mm/3%) and less than 1% in mean dose difference. The total calculation time is less than 2seconds, which allows for interactive isodose tuning. We developed a dose extension method that generates volumetric dose distribution from two surfaces. The application of dose extension is in interactive isodose tuning. The distance-based LUT fashion and correction strategy guarantee the computation efficiency and accuracy in isodose tuning.

Transferability of patients for radiation treatment between unmatched machines.

PurposeThe feasibility of transferring patients between unmatched machines for a limited number of treatment fractions was investigated for three‐dimensional conformal radiation therapy (3DCRT) and volumetric modulated arc therapy (VMAT) treatments.MethodsEighty patient‐plans were evaluated on two unmatched linacs: Elekta Versa HD and Elekta Infinity. Plans were equally divided into pelvis 3DCRT, prostate VMAT, brain VMAT, and lung VMAT plans. While maintaining the number of monitor units (MUs), plans were recalculated on the machine not originally used for treatment. Relative differences in dose were calculated between machines for the target volume and organs at risk (OARs). Differences in mean dose were assessed with paired t‐tests (p < 0.05). The number of interchangeable fractions allowable before surpassing a cumulative ±5% difference in dose was determined. Additionally, patient‐specific quality assurance (PSQA) measurements using ArcCHECK for both machines were compared with distributions calculated on the machine originally used for treatment using gradient compensation (GC) with 2%/2‐mm criteria.ResultsInterchanging the two machines for pelvic 3DCRT and VMAT (prostate, brain, and lung) plans resulted in an average change in target mean dose of 0.9%, −0.5%, 0.6%, 0.5%, respectively. Based on the differences in dose to the prescription point when changing machines, statistically, nearly one‐fourth of the prescribed fractions could be transferred between linacs for 3DCRT plans. While all of the prescribed fractions could typically be transferred among prostate VMAT plans, a rather large number of treatment fractions, 31% and 38%, could be transferred among brain and lung VMAT plans, respectively, without exceeding a ±5% change in the prescribed dose for two Elekta machines. Additionally, the OAR dosage was not affected within the given criterion with change of machine.ConclusionsDespite small differences in calculated dose, transferring patients between two unmatched Elekta machines with similar multileaf collimator (MLC)‐head for target coverage and minimum changes in OAR dose is possible for a limited number of fractions (≤3) to improve clinical flexibility and institutional throughput along with patient satisfaction. A similar study could be carried out for other machines for operational throughput.

Evaluation of patient-specific motion management for radiotherapy planning computed tomography using a statistical method.

We evaluated the probabilistic randomness of predictions by using individual numerical data based on general data for treatment planning computed tomography (CT) and evaluated the importance of patient-specific management through statistical analysis of our facility's data in lung stereotactic body radiotherapy (SBRT) and prostate volumetric modulated arc therapy (VMAT). The subjects were 30 patients who underwent lung SBRT with fiducial markers and 24 patients who underwent prostate VMAT. The average 3-dimensional (3D) displacement error between the fiducial marker and lung mass in 4DCT of lung SBRT was calculated and then compared with the 3D displacement error between the upper-lobe group (UG) and middle- or lower-lobe group (LG). The duty cycles between the lung tumor and fiducial marker at the <2-mm3 ambush area were compared between the UG and LG. In the prostate VMAT, the Shewhart control chart was analyzed by comparing multiple acquisition planning CT (MPCT) and cone-beam CT (CBCT) during the treatment period. The average 3D displacement errors in 4DCT for the lung tumor and fiducial marker were significantly different between the UG and middle- or lower-lobe group, but there was no correlation with the duty cycle. The Shewhart control chart for 3D displacement errors of the prostate for MPCT and CBCT showed that errors of >8 mm exceeded the control limit. In lung SBRT and prostate VMAT, overall statistical data from planning CT showed probabilistic randomness in predictions during the treatment period, and patient-specific motion management was needed to increase accuracy. A radiotherapy planning CT report showing a statistical analysis graph would be useful to objective share with staff.

Reduction of systematic dosimetric uncertainties in volumetric modulated arc therapy triggered by patient-specific quality assurance.

Background and purposeDosimetric patient-Specific Quality Assurance (PSQA) data contain in addition to cases with alerts, many cases without alerts. The aim of this study was to present a procedure to investigate long-term trend analysis of the complete set of PSQA data for the presence of site-specific deviations to reduce underlying systematic dose uncertainties.Materials and methodsThe procedure started by analysing a large set of prostate Volumetric Modulated Arc Therapy (VMAT) PSQA data obtained by comparing 3D electronic portal image device (EPID)_based in vivo dosimetry measurements with dose values predicted by the Treatment Planning System (TPS). If systematic deviations were present, several actions were required. These included confirmation of these deviations with an independent dose verification system for which a 2D detector array in a phantom was used, and analysing calculated with measured PSQA data, or delivery machine characteristics. Further analysis revealed that the under-dosage correlated with plan complexity and coincided with changes in clinically applied planning techniques.ResultsProstate VMAT PSQA data showed an under-dosage gradual increasing to about 2% in 3 years, which was confirmed by the measurements with the 2D detector array in a phantom. The implementation of new beam fits in the TPS led to a reduction of the observed deviations.ConclusionLong-term analysis of site-specific PSQA data is a useful method to monitor incremental changes in a radiotherapy department due to various changes in the treatment planning and delivery of prostate VMAT, and may lead to a reduction of systematic dose uncertainties in complex treatments.

Difference in VMAT dose distribution for prostate cancer with/without rectal gas removal and/or adaptive replanning.

We investigated differences in the volumetric-modulated arc therapy (VMAT) dose distribution in prostate cancer patients treated by rectal gas removal and/or adaptive replanning. Cone-beam computed tomography (CBCT) scans were performed daily for 22 treatments in eight prostate cancer patients with excessive rectal gas, and the CBCT images were analyzed. Rectal gas removal was performed, and irradiation was delivered after prostate matching. We compared dose-volume histograms for the daily CBCT images before and after rectal gas removal. Plan A was the original plan on CBCT images before rectal gas removal. Plan B was a single reoptimized plan on CBCT images before rectal gas removal. Plan C was the original plan on CBCT images after rectal gas removal. Plan D was a single reoptimized plan on CBCT images after rectal gas removal. D95 of the planning target volume (PTV) minus the rectum of Plan C (94.7% ± 6.6%) was significantly higher than that of Plan A (88.5% ± 10.4%). All dosimetric parameters of Plan C were improved by rectal gas removal compared with Plan A, regardless of the initial rectal gas volume. Dosimetric parameters of PTV minus the rectum of Plan B were significantly improved compared with Plan C. Additionally, the V78 of the rectal wall of Plan B (0.2% ± 0.5%) was significantly improved compared with Plan C (3.9% ± 6.3%, p = 0.003). The dosimetric parameters of Plan D were not significantly different from Plan B. The dose distribution of prostate VMAT was improved by rectal gas removal and/or adaptive replanning. An adaptive replanning on daily CBCT images might be a better method than rectal gas removal for prostate cancer patients with excessive rectal gas.

ORBIT-RT: A Real-Time, Open Platform for Knowledge-Based Quality Control of Radiotherapy Treatment Planning.

Access to knowledge-based treatment plan quality control has been hindered by the complexity of developing models and integration with different treatment planning systems (TPS). Online Real-time Benchmarking Information Technology for RadioTherapy (ORBIT-RT) provides a free, web-based platform for knowledge-based dose estimation that can be used by clinicians worldwide to benchmark the quality of their radiotherapy plans. The ORBIT-RT platform was developed to satisfy four primary design criteria: web-based access, TPS independence, Health Insurance Portability and Accountability Act compliance, and autonomous operation. ORBIT-RT uses a cloud-based server to automatically anonymize a user's Digital Imaging and Communications in Medicine for RadioTherapy (DICOM-RT) file before upload and processing of the case. From there, ORBIT-RT uses established knowledge-based dose-volume histogram (DVH) estimation methods to autonomously create DVH estimations for the uploaded DICOM-RT. ORBIT-RT performance was evaluated with an independent validation set of 45 volumetric modulated arc therapy prostate plans with two key metrics: (i) accuracy of the DVH estimations, as quantified by their error, DVHclinical - DVHprediction and (ii) time to process and display the DVH estimations on the ORBIT-RT platform. ORBIT-RT organ DVH predictions show < 1% bias and 3% error uncertainty at doses > 80% of prescription for the prostate validation set. The ORBIT-RT extensions require 3.0 seconds per organ to analyze. The DICOM upload, data transfer, and DVH output display extend the entire system workflow to 2.5-3 minutes. ORBIT-RT demonstrated fast and fully autonomous knowledge-based feedback on a web-based platform that takes only anonymized DICOM-RT as input. The ORBIT-RT system can be used for real-time quality control feedback that provides users with objective comparisons for final plan DVHs.

Predicting Three-Dimensional Dose Distribution of Prostate Volumetric Modulated Arc Therapy Using Deep Learning.

Background: Volumetric modulated arc therapy (VMAT) planning is a time-consuming process of radiation therapy. With a deep learning approach, 3D dose distribution can be predicted without the need for an actual dose calculation. This approach can accelerate the process by guiding and confirming the achievable dose distribution in order to reduce the replanning iterations while maintaining the plan quality. Methods: In this study, three dose distribution predictive models of VMAT for prostate cancer were developed, evaluated, and compared. Each model was designed with a different input data structure to train and test the model: (1) patient CT alone (PCT alone), (2) patient CT and generalized organ structure (PCTGOS), and (3) patient CT and specific organ structure (PCTSOS). The generative adversarial network (GAN) model was used as a core learning algorithm. The models were trained slice-by-slice using 46 VMAT plans for prostate cancer, and then used to predict and evaluate the dose distribution from 8 independent plans. Results: VMAT dose distribution was generated with a mean prediction time of approximately 3.5 s per patient, whereas the PCTSOS model was excluded due to a mean prediction time of approximately 17.5 s per patient. The highest average 3D gamma passing rate was 80.51 ± 5.94, while the lowest overall percentage difference of dose-volume histogram (DVH) parameters was 6.01 ± 5.44% for the prescription dose from the PCTGOS model. However, the PCTSOS model was the most reliable for the evaluation of multiple parameters. Conclusions: This dose prediction model could accelerate the iterative optimization process for the planning of VMAT treatment by guiding the planner with the desired dose distribution.

Error detection model developed using a multi‐task convolutional neural network in patient‐specific quality assurance for volumetric‐modulated arc therapy

In patient-specific quality assurance (QA) for static beam intensity-modulated radiation therapy (IMRT), machine-learning-based dose analysis methods have been developed to identify the cause of an error as an alternative to gamma analysis. Although these new methods have revealed that the cause of the error can be identified by analyzing the dose distribution obtained from the two-dimensional detector, they have not been extended to the analysis of volumetric-modulated arc therapy (VMAT) QA. In this study, we propose a deep learning approach to detect various types of errors in patient-specific VMAT QA. A total of 161 beams from 104 prostate VMAT plans were analyzed. All beams were measured using a cylindrical detector (Delta4; ScandiDos, Uppsala, Sweden), and predicted dose distributions in a cylindrical phantom were calculated using a treatment planning system (TPS). In addition to the error-free plan, we simulated 12 types of errors: two types of multileaf collimator positional errors (systematic or random leaf offset of 2mm), two types of monitor unit (MU) scaling errors (±3%), two types of gantry rotation errors (±2° in clockwise and counterclockwise direction), and six types of phantom setup errors (±1mm in lateral, longitudinal, and vertical directions). The error-introduced predicted dose distributions were created by editing the calculated dose distributions using a TPS with in-house software. Those 13 types of dose difference maps, consisting of an error-free map and 12 error maps, were created from the measured and predicted dose distributions and were used to train the convolutional neural network (CNN) model. Our model was a multi-task model that individually detected each of the 12 types of errors. Two datasets, Test sets 1 and 2, were prepared to evaluate the performance of the model. Test set 1 consisted of 13 types of dose maps used for training, whereas Test set 2 included the dose maps with 25 types of errors in addition to the error-free dose map. The dose map, which introduced 25 types of errors, was generated by combining two of the 12 types of simulated errors. For comparison with the performance of our model, gamma analysis was performed for Test sets 1 and 2 with the criteria set to 3%/2mm and 2%/1mm (dose difference/distance to agreement). For Test set 1, the overall accuracy of our CNN model, gamma analysis with the criteria set to 3%/2mm, and gamma analysis with the criteria set to 2%/1mm was 0.92, 0.19, and 0.81, respectively. Similarly, for Test set 2, the overall accuracy was 0.44, 0.42, and 0.95, respectively. Our model outperformed gamma analysis in the classification of dose maps containing a single type error, and the performance of our model was inferior in the classification of dose maps containing compound errors. A multi-task CNN model for detecting errors in patient-specific VMAT QA using a cylindrical measuring device was constructed, and its performance was evaluated. Our results demonstrate that our model was effective in identifying the error type in the dose map for VMAT QA.

Usability of detecting delivery errors during treatment of prostate VMAT with a gantry-mounted transmission detector.

Volumetric‐modulated arc therapy (VMAT) requires highly accurate control of multileaf collimator (MLC) movement, rotation speed of linear accelerator gantry, and monitor units during irradiation. Pretreatment validation and monitoring of these factors during irradiation are necessary for appropriate VMAT treatment. Recently, a gantry mounted transmission detector “Delta4 Discover® (D4D)” was developed to detect errors in delivering doses and dose distribution immediately after treatment. In this study, the performance of D4D was evaluated. Simulation plans, in which the MLC position was displaced by 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 mm from the clinically used original plans, were created for ten patients who received VMAT treatment for prostate cancer. Dose deviation (DD), distance‐to‐agreement (DTA), and gamma index analysis (GA) for each plan were evaluated by D4D. These results were compared to the results (DD, DTA and GA) measured by Delta4 Phantom + (D4P). We compared the deviations between the planned and measured values of the MLC stop positions A‐side and B‐side in five clinical cases of prostate VMAT during treatment and measured the GA values. For D4D, when the acceptable errors for DD, DTA, and GA were determined to be ≤3%, ≤2 mm, and ≤3%/2 mm, respectively, the minimum detectable errors in the MLC position were 2.0, 1.5, and 1.5 mm based on DD, DTA, and GA respectively. The corresponding minimum detectable MLC position errors were 2.0, 1.0, and 1.5 mm, respectively, for D4P. The deviation between the planned and measured position of MLC stopping point of prostate VMAT during treatment was stable at an average of −0.09 ± 0.05 mm, and all GA values were above 99.86%. In terms of delivering doses and dose distribution of VMAT, error detectability of D4D was comparable to that of D4P. The transmission‐type detector “D4D” is thus suitable for detecting delivery errors during irradiation.

A Novel Machine Learning Model for Dose Prediction in Prostate Volumetric Modulated Arc Therapy Using Output Initialization and Optimization Priorities.

Treatment planning for prostate volumetric modulated arc therapy (VMAT) can take 5–30 min per plan to optimize and calculate, limiting the number of plan options that can be explored before the final plan decision. Inspired by the speed and accuracy of modern machine learning models, such as residual networks, we hypothesized that it was possible to use a machine learning model to bypass the time-intensive dose optimization and dose calculation steps, arriving directly at an estimate of the resulting dose distribution for use in multi-criteria optimization (MCO). In this study, we present a novel machine learning model for predicting the dose distribution for a given patient with a given set of optimization priorities. Our model innovates upon the existing machine learning techniques by utilizing optimization priorities and our understanding of dose map shapes to initialize the dose distribution before dose refinement via a voxel-wise residual network. Each block of the residual network individually updates the initialized dose map before passing to the next block. Our model also utilizes contiguous and atrous patch sampling to effectively increase the receptive fields of each layer in the residual network, decreasing its number of layers, increasing model prediction and training speed, and discouraging overfitting without compromising on the accuracy. For analysis, 100 prostate VMAT cases were used to train and test the model. The model was evaluated by the training and testing errors produced by 50 iterations of 10-fold cross-validation, with 100 cases randomly shuffled into the subsets at each iteration. The error of the model is modest for this data, with average dose map root-mean-square errors (RMSEs) of 2.38 ± 0.47% of prescription dose overall patients and all optimization priority combinations in the patient testing sets. The model was also evaluated at iteratively smaller training set sizes, suggesting that the model requires between 60 and 90 patients for optimal performance. This model may be used for quickly estimating the Pareto set of feasible dose objectives, which may directly accelerate the treatment planning process and indirectly improve final plan quality by allowing more time for plan refinement.

A practical method to quantify knowledge-based DVH prediction accuracy and uncertainty with reference cohorts.

The adoption of knowledge‐based dose‐volume histogram (DVH) prediction models for assessing organ‐at‐risk (OAR) sparing in radiotherapy necessitates quantification of prediction accuracy and uncertainty. Moreover, DVH prediction error bands should be readily interpretable as confidence intervals in which to find a percentage of clinically acceptable DVHs. In the event such DVH error bands are not available, we present an independent error quantification methodology using a local reference cohort of high‐quality treatment plans, and apply it to two DVH prediction models, ORBIT‐RT and RapidPlan, trained on the same set of 90 volumetric modulated arc therapy (VMAT) plans. Organ‐at‐risk DVH predictions from each model were then generated for a separate set of 45 prostate VMAT plans. Dose‐volume histogram predictions were then compared to their analogous clinical DVHs to define prediction errors Vclin,i‐Vpred,i (ith plan), from which prediction bias μ, prediction error variation σ, and root‐mean‐square error RMSEpred≡1N∑iVclin,i‐Vpred,i2≅σ2+μ2 could be calculated for the cohort. The empirical RMSEpred was then contrasted to the model‐provided DVH error estimates. For all prostate OARs, above 50% Rx dose, ORBIT‐RT μ and σ were comparable to or less than those of RapidPlan. Above 80% Rx dose, μ < 1% and σ < 3‐4% for both models. As a result, above 50% Rx dose, ORBIT‐RT RMSEpred was below that of RapidPlan, indicating slightly improved accuracy in this cohort. Because μ ≈ 0, RMSEpred is readily interpretable as a canonical standard deviation σ, whose error band is expected to correctly predict 68% of normally distributed clinical DVHs. By contrast, RapidPlan's provided error band, although described in literature as a standard deviation range, was slightly less predictive than RMSEpred (55–70% success), while the provided ORBIT‐RT error band was confirmed to resemble an interquartile range (40–65% success) as described. Clinicians can apply this methodology using their own institutions’ reference cohorts to (a) independently assess a knowledge‐based model's predictive accuracy of local treatment plans, and (b) interpret from any error band whether further OAR dose sparing is likely attainable.