Prediction of VMAT delivery accuracy with textural features calculated from fluence maps

Verifying institutionally developed hybrid 3D-printed coaxial cylindrical phantom for patient-specific quality assurance in stereotactic body radiation therapy of hepatocellular carcinoma

An accurate and reliable patient-specific quality assurance (PSQA) is crucial to ensure the safety and precision of Stereotactic body radiation therapy (SBRT) in treating Hepatocellular carcinoma (HCC). This study examines the effectiveness of a novel hybrid 3D-printed hybrid coaxial cylindrical phantom for PSQA in the SBRT of HCC. The study compared three different point dose verification techniques for PSQA: a traditional solid water phantom, two dimensional detector array I'MatriXX, and a newly developed hybrid 3D-printed phantom. Thirty SBRT HCC liver cases were examined using these techniques, and point doses were measured and compared to planned doses using the perpendicular composite method with solid water and I'MatriXX phantoms. Unlike the other two methods, the point dose was compared in true composite geometry using the hybrid 3D-printed phantom, which enhanced the accuracy and consistency of PSQA. The study aims to assess the statistical significance and accuracy of the hybrid 3D-printed phantom compared to other methods. The results showed all techniques complied with the institutional threshold criteria of within ± 3% for point-dose measurement discrepancies. The hybrid 3D-printed phantom was found to have better consistency with a lower standard deviation than traditional methods. Statistical analysis using Student's t-test revealed the statistical significance of the hybrid 3D-printed phantom technique in patient-specific point-dose assessments with a p-value < 0.01. The hybrid 3D-printed phantom developed institutionally is cost-effective and easy to handle. It has been proven to be a valuable tool for PSQA in SBRT for the treatment of HCC and has demonstrated its practicality and reliability.

Prediction of patient-specific quality assurance for volumetric modulated arc therapy using radiomics-based machine learning with dose distribution

PURPOSE

We sought to develop machine learning models to predict the results of patient-specific quality assurance (QA) for volumetric modulated arc therapy (VMAT), which were represented by several dose-evaluation metrics-including the gamma passing rates (GPRs)-and criteria based on the radiomic features of 3D dose distribution in a phantom.

METHODS

A total of 4,250 radiomic features of 3D dose distribution in a cylindrical dummy phantom for 140 arcs from 106 clinical VMAT plans were extracted. We obtained the following dose-evaluation metrics: GPRs with global and local normalization, the dose difference (DD) in 1% and 2% passing rates (DD1% and DD2%) for 10% and 50% dose threshold, and the distance-to-agreement in 1-mm and 2-mm passing rates (DTA1 mm and DTA2 mm) for 0.5%/mm and 1.0%.mm dose gradient threshold determined by measurement using a diode array in patient-specific QA. The machine learning regression models for predicting the values of the dose-evaluation metrics using the radiomic features were developed based on the elastic net (EN) and extra trees (ET) models. The feature selection and tuning of hyperparameters were performed with nested cross-validation in which four-fold cross-validation is used within the inner loop, and the performance of each model was evaluated in terms of the root mean square error (RMSE), the mean absolute error (MAE), and Spearman's rank correlation coefficient.

RESULTS

The RMSE and MAE for the developed machine learning models ranged from <1% to nearly <10% depending on the dose-evaluation metric, the criteria, and dose and dose gradient thresholds used for both machine learning models. It was advantageous to focus on high dose region for predicating global GPR, DDs, and DTAs. For certain metrics and criteria, it was possible to create models applicable for patients' heterogeneity by training only with dose distributions in phantom.

CONCLUSIONS

The developed machine learning models showed high performance for predicting dose-evaluation metrics especially for high dose region depending on the metric and criteria. Our results demonstrate that the radiomic features of dose distribution can be considered good indicators of the plan complexity and useful in predicting measured dose evaluation metrics.

Progressive resolution optimizer (PRO) predominates over photon optimizer (PO) in sparing of spinal cord for spine SABR VMAT plans

BACKGROUND

we assessed the performance of the optimization algorithms by comparing volumetric modulated arc therapy generated by a progressive resolution optimized (VMATPRO) and photon optimizer (VMATPO) in terms of plan quality, MU reduction, sparing of the spinal cord (or cauda equina), and plan complexity.

METHODS

Fifty-seven patients who received spine stereotactic ablative radiotherapy (SABR) with tumors located in the cervical, thoracic, and lumbar spine were retrospectively selected. For each patient, VMATPRO and VMATPO with two full arcs were generated with using the PRO and PO algorithms. For dosimetric evaluation, the dose-volumetric (DV) parameters of the planning target volume (PTV), organs at risk (OARs), the corresponding planning organs at risk (PRV), and 1.5-cm ring structure surrounding the PTV (Ring1.5 cm) were calculated for all VMAT plans. The total number of monitor units (MUs) and the modulation complexity score for the VMAT (MCSv) were compared. To investigate the correlations of OAR sparing to plan complexity, Pearson's and Spearman's correlation tests were conducted between the two algorithms (PO - PRO, denoted as Δ) in the DV parameters for normal tissues, total MUs, and MCSv.

RESULTS

For the PTVs, Target conformity and dose homogeneity in the PTVs of VMATPRO were better than those of VMATPO with statistical significance. For the spinal cords (or cauda equine) and the corresponding PRVs, all of the DV parameters for VMATPRO were markedly lower than those for VMATPO, with statistical significance (all p < 0.0001). Among them, the difference in the maximum dose to the spinal cord between VMATPRO and VMATPO was remarkable (9.04 Gy vs. 11.08 Gy with p < 0.0001). For Ring1.5 cm, no significant difference in V115% for VMATPRO and VMATPO was observed.

CONCLUSIONS

The use of VMATPRO resulted in improved coverage and uniformity of dose to the PTV, as well as OARs sparing, compared with that of VMATPO for cervical, thoracic, and lumbar spine SABR. Better dosimetric plan quality generated by the PRO algorithm was observed to result in higher total MUs and plan complexity. Therefore, careful evaluation of its deliverability should be performed with caution during the routine use of the PRO algorithm.

Comparison of dosimetric effects of MLC positional errors on VMAT and IMRT plans for SBRT radiotherapy in non-small cell lung cancer

The positional accuracy of multi-leaf collimators (MLC) is important in stereotactic body radiotherapy (SBRT). The aim of this study was to investigate the impact between MLC positional error and dosimetry of volume intensity modulated (VMAT) and general intensity modulated (IMRT) plans for non-small cell lung cancer (NSCLC). Fifteen patients with NSCLC were selected to design the 360 SBRT-VMAT plans and the 360 SBRT-IMRT error plans. The DICOM files for these treatment plans were imported into a proprietary computer program that introduced delivery errors. Random and systematic MLC position (0.1, 0.2, 0.5, 1.0, 1.5, and 2.0 mm) errors were introduced. The systematic errors were shift errors (caused by gravity), opening errors, and closing errors. The CI, GI, d2cm and generalized equivalent uniform dose (gEUD) were calculated for the original plan and all treatment plans, accounting for the errors. Dose sensitivity was calculated using linear regression for MLC position errors. The random MLC errors were relatively insignificant. MLC shift, opening, and closing errors had a significant effect on the dose distribution of the SBRT plan. VMAT was more significant than IMRT. To ensure that the gEUD variation of PTV is controlled within 2%, the shift error, opening error, and closing error of IMRT should be less than 2.4 mm, 1.15 mm, and 0.97 mm, respectively. For VMAT, the shift error, opening error, and closing error should be less than 0.95 mm, 0.32 mm, and 0.38 mm, respectively. The dose sensitivity results obtained in this study can be used as a guide for patient-based quality assurance efforts. The position error of the MLC system had a significant impact on the gEUD of the SBRT technology. The MLC systematic error has a greater dosimetric impact on the VMAT plan than on the IMRT plan for SBRT, which should be carefully monitored.

Radiomics-guided radiation therapy: opportunities and challenges

Radiomics is an advanced image-processing framework, which extracts image features and considers them as biomarkers towards personalized medicine. Applications include disease detection, diagnosis, prognosis, and therapy response assessment/prediction. As radiation therapy aims for further individualized treatments, radiomics could play a critical role in various steps before, during and after treatment. Elucidation of the concept of radiomics-guided radiation therapy (RGRT) is the aim of this review, attempting to highlight opportunities and challenges underlying the use of radiomics to guide clinicians and physicists towards more effective radiation treatments. This work identifies the value of RGRT in various steps of radiotherapy from patient selection to follow-up, and subsequently provides recommendations to improve future radiotherapy using quantitative imaging features.

Machine learning models to predict the delivered positions of Elekta multileaf collimator leaves for volumetric modulated arc therapy

PURPOSE

Accurate positioning of multileaf collimator (MLC) leaves during volumetric modulated arc therapy (VMAT) is essential for accurate treatment delivery. We developed a linear regression, support vector machine, random forest, extreme gradient boosting (XGBoost), and an artificial neural network (ANN) for predicting the delivered leaf positions for VMAT plans.

METHODS

For this study, 160 MLC log files from 80 VMAT plans were obtained from a single institution treated on 3 Elekta Versa HD linear accelerators. The gravity vector, X1 and X2 jaw positions, leaf gap, leaf position, leaf velocity, and leaf acceleration were extracted and used as model inputs. The models were trained using 70% of the log files and tested on the remaining 30%. Mean absolute error (MAE), root mean square error (RMSE), the coefficient of determination R² , and fitted line plots showing the relationship between delivered and predicted leaf positions were used to evaluate model performance.

RESULTS

The models achieved the following errors: linear regression (MAE = 0.158 mm, RMSE = 0.225 mm), support vector machine (MAE = 0.141 mm, RMSE = 0.199 mm), random forest (MAE = 0.161 mm, RMSE = 0.229 mm), XGBoost (MAE = 0.185 mm, RMSE = 0.273 mm), and ANN (MAE = 0.361 mm, RMSE = 0.521 mm). A significant correlation between a plan's gamma passing rate (GPR) and the prediction errors of linear regression, support vector machine, and random forest is seen (p < 0.045).

CONCLUSIONS

We examined various models to predict the delivered MLC positions for VMAT plans treated with Elekta linacs. Linear regression, support vector machine, random forest, and XGBoost achieved lower errors than ANN. Models that can accurately predict the individual leaf positions during treatment can help identify leaves that are deviating from the planned position, which can improve a plan's GPR.

Predicting gamma evaluation results of patient-specific head and neck volumetric-modulated arc therapy quality assurance based on multileaf collimator patterns and fluence map features: A feasibility study

The purpose of this study was to develop a predictive model for patient-specific VMAT QA results using multileaf collimator (MLC) effect and texture analysis. The MLC speed, acceleration and texture analysis features were extracted from 106 VMAT plans as predictors. Gamma passing rate (GPR) was collected as a response class with gamma criteria of 2%/2 mm and 3%/2 mm. The model was trained using two machine learning methods: AdaBoost classification and bagged regression trees model. GPR was classified into the "PASS" and "FAIL" for the classification model using the institutional warning level. The accuracy of the model was assessed using sensitivity and specificity. In addition, the accuracy of the regression model was determined using the difference between predicted and measured GPR. For the AdaBoost classification model, the sensitivity/specificity was 94.12%/100% and 63.63%/53.13% at gamma criteria of 2%/2 mm and 3%/2 mm, respectively. For the bagged regression trees model, the sensitivity/specificity was 94.12%/91.89% and 61.18%/68.75% at gamma criteria of 2%/2 mm and 3%/2 mm, respectively. The root mean square error (RMSE) of difference between predicted and measured GPR was found at 2.44 and 1.22 for gamma criteria of 2%/2 mm and 3%/2 mm, respectively. The promising result was found at tighter gamma criteria 2%/2 mm with 94.12% sensitivity (both bagged regression trees and AdaBoost classification model) and 100% specificity (AdaBoost classification model).

Improvement Using Planomics Features on Prediction and Classification of Patient-Specific Quality Assurance Using Head and Neck Volumetric Modulated Arc Therapy Plan

Purpose: This study aimed to evaluate the utility of a new plan feature (planomics feature) for predicting the results of patient-specific quality assurance using the head and neck (H&N) volumetric modulated arc therapy (VMAT) plan. Methods: One hundred and thirty-one H&N VMAT plans in our institution from 2019 to 2021 were retrospectively collected. Dosimetric verification for all plans was carried out using the portal dosimetry system integrated into the Eclipse treatment planning system based on the electronic portal imaging devices. Gamma passing rates (GPR) were analyzed using three gamma indices of 3%/3 mm, 3%/2 mm, and 2%/2 mm with a 10% dose threshold. Forty-eight conventional features affecting the dose delivery accuracy were used in the study, and 2,476 planomics features were extracted based on the radiotherapy plan file. Three prediction and classification models using conventional features (CF), planomics features (PF), and hybrid features (HF) combining two sets of features were constructed by the gradient boosting regressor (GBR) and Ridge classifier for each GPR of 3%/3 mm, 3%/2 mm, and 2%/2 mm, respectively. The absolute prediction error (APE) and the area under the curve (AUC) were adopted for assessing the performance of prediction and classification models. Results: In the GPR prediction, the average APE of the models using CF, PF, and HF was 1.3 ± 1.2%/3.6 ± 3.0%, 1.7 ± 1.5%/3.8 ± 3.5%, and 1.1 ± 1.0%/4.1 ± 3.1% for 2%/2 mm; 0.7 ± 0.6%/2.0 ± 2.0%, 1.0±1.1%/2.2 ± 1.8%, and 0.6 ± 0.6%/2.2 ± 1.9% for 3%/2 mm; and 0.4 ± 0.3%/1.2 ± 1.2%, 0.4±0.5%/1.3 ± 1.0%, and 0.3±0.3%/1.2 ± 1.1% for 3%/3 mm, respectively. In the regression prediction, three models give a similar modeling performance for predicting the GPR. The classification results were 0.67 ± 0.03/0.66 ± 0.07, 0.77 ± 0.03/0.73 ± 0.06, and 0.78 ± 0.02/0.75 ± 0.04 for 3%/3 mm, respectively. For 3%/2 mm, the AUCs of the training and testing cohorts were 0.64 ± 0.03/0.62 ± 0.07, 0.70 ± 0.03/0.67 ± 0.06, and 0.75 ± 0.03/0.71 ± 0.07, respectively, and for 2%/2 mm, the average AUCs of the training and testing cohorts were 0.72 ± 0.03/0.72 ± 0.06, 0.78 ± 0.04/0.73 ± 0.07, and 0.81 ± 0.03/0.75 ± 0.06, respectively. In the classification, the PF model has a better classification performance than the CF model. Moreover, the HF model provides the best result among the three classifications models. Conclusions: The planomics features can be used for predicting and classifying the GPR results and for improving the model performance after combining the conventional features for the GPR classification.

On the use of trajectory log files for machine & patient specific QA

Purpose:Trajectory log files are increasingly being utilized clinically for machine and patient specific QA. The process of converting the DICOM-RT plan to a deliverable trajectory by the linac control software introduces some uncertainty that is inherently incorporated into measurement-based patient specific QA but is not necessarily included for trajectory log file-based methods. Roughly half of prior studies have included this uncertainty in the analysis while the remaining studies have ignored it, and it has yet to be quantified in the literature.Methods:We collected DICOM-RT files from the treatment planning system and the trajectory log files from four TrueBeam linear accelerators for 25 IMRT and 10 VMAT plans. We quantified the DICOM-RT Conversion to Trajectory Residual (DCTR, difference between 'planned' MLC position from TPS DICOM-RT file and 'expected' MLC position (the deliverable MLC positions calculated by the linac control software) from trajectory log file) and compared it to the discrepancy between actual and expected machine parameters recorded in trajectory log files.Results:RMS of the DCTR was 0.0845 mm (range of RMS per field/arc: 0.0173-0.1825 mm) for 35 plans (114 fields/arcs) and was independent of treatment technique, with a maximum observed discrepancy at any control point of 0.7255 mm. DCTR was correlated with MLC velocity and was consistent over the course of treatment and over time, with a slight change in magnitude observed after a linac software upgrade. For comparison, the RMS of trajectory log file reported delivery error for moving MLCs was 0.0205 mm, thus DCTR is about four times the recorded delivery error in the trajectory log file.Conclusion:The uncertainty introduced from the conversion process by the linac control software from DICOM-RT plan to a deliverable trajectory is 3-4 times larger than the discrepancy between actual and expected machine parameters recorded in trajectory log files. This uncertainty should be incorporated into the analysis when using trajectory log file-based methods for analyzing MLC performance or patient-specific QA.

Patient-specific quality assurance using machine log files analysis for stereotactic body radiation therapy (SBRT)

An in‐house trajectory log analysis program (LOGQA) was developed to evaluate the delivery accuracy of volumetric‐modulated arc therapy (VMAT) for stereotactic body radiation therapy (SBRT). Methods have been established in LOGQA to provide analysis on dose indices, gantry angles, and multi‐leaf collimator (MLC) positions. Between March 2019 and May 2020, 120 VMAT SBRT plans of various treatment sites using flattening filter‐free (FFF) mode were evaluated using both LOGQA and phantom measurements. Gantry angles, dose indices, and MLC positions were extracted from log and compared with each plan. Integrated transient fluence map (ITFM) was reconstructed from log to examine the deviation of delivered fluence against the planned one. Average correlation coefficient of dose index versus gantry angle and ITFM for all patients were 1.0000, indicating that the delivered beam parameters were in good agreement with planned values. Maximum deviation of gantry angles and monitor units (MU) of all patients were less than 0.2 degree and 0.03 % respectively. Regarding MLC positions, maximum and root‐mean‐square (RMS) deviations from planned values were less than 0.6 mm and 0.3 mm respectively, indicating that MLC positions during delivery followed planned values in precise manner. Results of LOGQA were consistent with measurement, where all gamma‐index passing rates were larger than 95 %, with 2 %/2 mm criteria. Three types of intentional errors were introduced to patient plan for software validation. LOGQA was found to recognize the introduced errors of MLC positions, gantry angles, and dose indices with magnitudes of 1 mm, 1 degree, and 5 %, respectively, which were masked in phantom measurement. LOGQA was demonstrated to have the potential to reduce or even replace patient‐specific QA measurements for SBRT plan delivery provided that the frequency and amount of measurement‐based machine‐specific QA can be increased to ensure the log files record real values of machine parameters.

Improvement of prediction and classification performance for gamma passing rate by using plan complexity and dosiomics features

PURPOSE

The purpose of this study was to predict and classify the gamma passing rate (GPR) value by using new features (3D dosiomics features and combined with plan and dosiomics features) together with a machine learning technique for volumetric modulated arc therapy (VMAT) treatment plans.

METHODS AND MATERIALS

A total of 888 patients who underwent VMAT were enrolled comprising 1255 treatment plans. Further, 24 plan complexity features and 851 dosiomics features were extracted from the treatment plans. The dataset was randomly split into a training/validation (80%) and test (20%) dataset. The three models for prediction and classification using XGBoost were as follows: (i) plan complexity features-based prediction method (plan model); (ii) 3D dosiomics feature-based prediction model (dosiomics model); (iii) a combination of both the previous models (hybrid model). The prediction performance was evaluated by calculating the mean absolute error (MAE) and the correlation coefficient (CC) between the predicted and measured GPRs. The classification performance was evaluated by calculating the area under curve (AUC) and sensitivity.

RESULTS

MAE and CC at γ2%/2 mm in the test dataset were 4.6% and 0.58, 4.3% and 0.61, and 4.2% and 0.63 for the plan model, dosiomics model, and hybrid model, respectively. AUC and sensitivity at γ2%/2 mm in test dataset were 0.73 and 0.70, 0.81 and 0.90, and 0.83 and 0.90 for the plan model, dosiomics model, and hybrid model, respectively.

CONCLUSIONS

A combination of both plan and dosiomics features with machine learning technique can improve the prediction and classification performance for GPR.