Predicting the error magnitude in patient-specific QA during radiotherapy based on ResNet.

BACKGROUND: The error magnitude is closely related to patient-specific dosimetry and plays an important role in evaluating the delivery of the radiotherapy plan in QA. No previous study has investigated the feasibility of deep learning to predict error magnitude. OBJECTIVE: The purpose of this study was to predict the error magnitude of different delivery error types in radiotherapy based on ResNet. METHODS: A total of 34 chest cancer plans (172 fields) of intensity-modulated radiation therapy (IMRT) from Eclipse were selected, of which 30 plans (151 fields) were used for model training and validation, and 4 plans including 21 fields were used for external testing. The collimator misalignment (COLL), monitor unit variation (MU), random multi-leaf collimator shift (MLCR), and systematic MLC shift (MLCS) were introduced. These dose distributions of portal dose predictions for the original plans were defined as the reference dose distribution (RDD), while those for the error-introduced plans were defined as the error-introduced dose distribution (EDD). Different inputs were used in the ResNet for predicting the error magnitude. RESULTS: In the test set, the accuracy of error type prediction based on the dose difference, gamma distribution, and RDD + EDD was 98.36%, 98.91%, and 100%, respectively; the root mean squared error (RMSE) was 1.45-1.54, 0.58-0.90, 0.32-0.36, and 0.15-0.24; the mean absolute error (MAE) was 1.06-1.18, 0.32-0.78, 0.25-0.27, and 0.11-0.18, respectively, for COLL, MU, MLCR and MLCS. CONCLUSIONS: In this study, error magnitude prediction models with dose difference, gamma distribution, and RDD + EDD are established based on ResNet. The accurate prediction of the error magnitude under different error types can provide reference for error analysis in patient-specific QA.

Improvement of Prediction Performance for Radiation Pneumonitis by Using 3-Dimensional Dosiomic Features.

INTRODUCTION: Patients with early non-small-cell lung cancer (NSCLC) have a relatively long survival time after stereotactic body radiation therapy (SBRT). Predicting radiation-induced pneumonia (RP) has important clinical and social implications for improving the quality of life of such patients. This study developed an RP prediction model by using 3-dimensional (3D) dosiomic features. The model can be used to guide radiation therapy to reduce toxicity. METHODS: Radiomic features were extracted from pre-treatment CT, dose-volume histogram (DVH) parameters and dosiomic features were extracted from the 3D dose distribution of 140 lung cancer patients. Four predictive models: (1) CT; (2) CT + DVH; (3) CT + Rtdose; and (4) Hybrid, CT + DVH + Rtdose, were trained to predict symptomatic RP by extremely randomized trees. Accuracy, sensitivity, specificity, and area under the receiver operator characteristic curve were evaluated. RESULT: Results showed that the fraction regimen was correlated with symptomatic RP (P < .001). The proposed model achieved promising prediction results. The performance metrics for CT, CT + DVH, CT + Rtdose, and Hybrid were as follows: accuracy: 0.786, 0.821, 0.821, and 0.857; sensitivity: 0.625, 1, 0.875, and 1; specificity: 0.8, 0.565, 0.5, and 0.875; and area under the receiver operator characteristic curve: 0.791, 0.809, 0.907, and 0.920, respectively. CONCLUSION: Dosiomic features can improve the performance of the predictive model for symptomatic RP compared with that obtained with the CT + DVH model. The model proposed in this study can help radiation oncologists individually predict the incidence rate of RP.

Dosimetry and treatment efficiency of SBRT using TaiChiB radiotherapy system for two-lung lesions with one overlapping organs at risk.

Purpose: This study aims to assess the dosimetry and treatment efficiency of TaiChiB-based Stereotactic Body Radiotherapy (SBRT) plans applying to treat two-lung lesions with one overlapping organs at risk. Methods: For four retrospective patients diagnosed with two-lung lesions each patient, four treatment plans were designed including Plan Edge, TaiChiB linac-based, RGS-based, and a linac-RGS hybrid (Plan TCLinac, Plan TCRGS, and Plan TCHybrid). Dosimetric metrics and beam-on time were employed to evaluate and compare the TaiChiB-based plans against Plan Edge. Results: For Conformity Index (CI), Plan TCRGS outperformed all other plans with an average CI of 1.06, as opposed to Plan Edge's 1.33. Similarly, for R50 %, Plan TCRGS was superior with an average R50 % of 3.79, better than Plan Edge's 4.28. In terms of D2 cm, Plan TCRGS also led with an average of 48.48%, compared to Plan Edge's 56.25%. For organ at risk (OAR) sparing, Plan TCRGS often displayed the lowest dosimetric values, notably for the spinal cord (Dmax 5.92 Gy) and lungs (D1500cc 1.00 Gy, D1000cc 2.61 Gy, V10 Gy 15.14%). However, its high Dmax values for the heart and great vessels sometimes exceeded safety thresholds. Plan TCHybrid presented a balanced approach, showing doses comparable to or better than Plan Edge without crossing safety limits. In terms of beam-on time, Plan TCLinac emerged as the most efficient treatment option in three out of four cases, followed closely by Plan Edge in one case. Plan TCRGS, despite its dosimetric advantages, was the least efficient, recording notably longer beam-on times, with a peak at 33.28 minutes in Case 2. Conclusion: For patients with two-lung lesions treated by SBRT whose one lesion overlaps with OARs, the Plan TCHybrid delivered by TaiChiB digital radiotherapy system can be recommended as a clinical option.

AS-NeSt: A Novel 3D Deep Learning Model for Radiation Therapy Dose Distribution Prediction in Esophageal Cancer Treatment With Multiple Prescriptions.

PURPOSE: Implementing artificial intelligence technologies allows for the accurate prediction of radiation therapy dose distributions, enhancing treatment planning efficiency. However, esophageal cancers present unique challenges because of tumor complexity and diverse prescription types. Additionally, limited data availability hampers the effectiveness of existing artificial intelligence models. This study developed a deep learning model, trained on a diverse data set of esophageal cancer prescriptions, to improve dose prediction accuracy. METHODS AND MATERIALS: We retrospectively collected data from 530 patients with esophageal cancer, including single-target and simultaneous integrated boost prescriptions, for model building. The proposed Asymmetric ResNeSt (AS-NeSt) model features novel 3-dimensional (3D) ResNeSt blocks and an asymmetrical architecture. We constructed a loss function targeting global and local doses and validated the model's performance against existing alternatives. Model-assisted experiments were used to validate its clinical benefits. RESULTS: The AS-NeSt model maintained an absolute prediction error below 5% for each dosimetric metric. The average Dice similarity coefficient for isodose volumes was 0.93. The model achieved an average relative prediction error of 2.02%, statistically lower than Hierarchically Densely Connected U-net (4.17%), DoseNet (2.35%), and Densely Connected Network (3.65%). It also demonstrated significantly fewer parameters and shorter prediction times. Clinically, the AS-NeSt model raised physicians' ability to accurately preassess appropriate treatment methods before planning from 95.24% to 100%, reduced planning time by over 61% for junior dosimetrists and 52% for senior dosimetrists, and decreased both inter- and intra-dosimetrist discrepancies by more than 50%. CONCLUSIONS: The AS-NeSt model, developed with innovative 3D ResNeSt blocks and an asymmetrical encoder-decoder structure, has been validated using clinical esophageal cancer patient data. It accurately predicts 3D dose distributions for various prescriptions, including simultaneous integrated boost, showing potential to improve the management of esophageal cancer treatment in a clinical setting.

A planning strategy may reduce the risk of heart diseases and radiation pneumonia: Avoiding the specific heart substructures.

BACKGROUND: Dose to heart substructures is a better predictor for major adverse cardiac events (MACE) than mean heart dose (MHD). We propose an avoidance planning strategy for important cardiac substructures. MATERIAL AND METHODS: Two plans, clinical and cardiac substructure-avoidance plan, were generated for twenty patients. Five dose-sensitive substructures, including left ventricle, pulmonary artery, left anterior descending branch, left circumflex branch and the coronary artery were chosen. The avoidance plan aims to meet the target criteria and organ-at-risk (OARs) constraints while minimizing the dose parameters of the above five substructures. The dosimetric assessments included the mean dose and the maximum dose of cardiac substructures and several volume parameters. In addition, we also evaluated the relative risk of coronary artery disease (CAD), chronic heart failure (CHF), and radiation pneumonia (RP). RESULTS: Pearson correlation coefficient and R2 value of linear regression fitting demonstrated that MHD had poor prediction ability for the mean dose of the cardiac substructures. Compared to clinical plans, an avoidance plan is able to statistically significantly decrease the dose to key substructures. Meanwhile, the dose to OARs and the coverage of the target are comparable in the two plans. In addition, it can be observed that the avoidance plan statistically decreases the relative risks of CAD, CHF, and RP. CONCLUSIONS: The substructure-avoidance planning strategy that incorporates the cardiac substructures into optimization process, can protect the important heart substructures, such as left ventricle, left anterior descending branch and pulmonary artery, achieving the substantive sparing of dose-sensitive cardiac structures, and have the potential to decrease the relative risks of CAD, CHF, and RP.

Optimization of isocenter position for multiple brain metastases single-isocenter stereotactic radiosurgery to minimize dosimetric variations due to rotational uncertainty.

PURPOSE: This paper studied a novel calculation framework that can determine the optimal value isocenter position of single isocenter SRS treatment plan for multiple brain metastases, in order to minimize the dosimetric variations caused by rotational uncertainty. MATERIALS AND METHODS: 21 patients with 2-4 GTVswho received SRS treatment for multiple brain metastases in our institution were selected for the retrospective study. The PTVwas obtained by expanding GTV 1 mm isotropic margin. We studied a stochastic optimization framework, which determined the optimal value isocenter location by maximizing the average target dose coverageCtarget,meanwith a rotation error of no more than 1°. We evaluated the performance of the optimal isocenter by comparing theCtarget,meanand average dice similarity coefficient (DSC)with the optimal value and the center of mass (CM) respectively as the treatment isocenter. The extra PTV margin to achieve 100% target dose coverage was calculated by our framework. RESULTS: Compared to the CM method, the optimal value isocenter method increased the averageCtarget,meanof all targets from 97.0% to 97.7%and the average DSC from 0.794to 0.799. Throughout all the cases, the average extra PTV margin to obtain full target dose coverage was 0.7 mmwhen using the optimal value isocenter as the treatment isocenter. CONCLUSION: We studied a novel computational framework using stochastic optimization to determine the optimal isocenter position of SRS treatment plan for multiple brain metastases. At the same time, our framework gave the extra PTV margin to obtain full target dose coverage.

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

OBJECTIVE: To identify delivery error type and predict associated error magnitude by image-based features using machine learning (ML). METHODS: In 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. RESULTS: The 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. CONCLUSION: It 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.

Improving CBCT image quality to the CT level using RegGAN in esophageal cancer adaptive radiotherapy.

OBJECTIVE: This study aimed to improve the image quality and CT Hounsfield unit accuracy of daily cone-beam computed tomography (CBCT) using registration generative adversarial networks (RegGAN) and apply synthetic CT (sCT) images to dose calculations in radiotherapy. METHODS: The CBCT/planning CT images of 150 esophageal cancer patients undergoing radiotherapy were used for training (120 patients) and testing (30 patients). An unsupervised deep-learning method, the 2.5D RegGAN model with an adaptively trained registration network, was proposed, through which sCT images were generated. The quality of deep-learning-generated sCT images was quantitatively compared to the reference deformed CT (dCT) image using mean absolute error (MAE), root mean square error (RMSE) of Hounsfield units (HU), and peak signal-to-noise ratio (PSNR). The dose calculation accuracy was further evaluated for esophageal cancer radiotherapy plans, and the same plans were calculated on dCT, CBCT, and sCT images. RESULTS: The quality of sCT images produced by RegGAN was significantly improved compared to the original CBCT images. ReGAN achieved image quality in the testing patients with MAE sCT vs. CBCT: 43.7 ± 4.8 vs. 80.1 ± 9.1; RMSE sCT vs. CBCT: 67.2 ± 12.4 vs. 124.2 ± 21.8; and PSNR sCT vs. CBCT: 27.9 ± 5.6 vs. 21.3 ± 4.2. The sCT images generated by the RegGAN model showed superior accuracy on dose calculation, with higher gamma passing rates (93.3 ± 4.4, 90.4 ± 5.2, and 84.3 ± 6.6) compared to original CBCT images (89.6 ± 5.7, 85.7 ± 6.9, and 72.5 ± 12.5) under the criteria of 3 mm/3%, 2 mm/2%, and 1 mm/1%, respectively. CONCLUSION: The proposed deep-learning RegGAN model seems promising for generation of high-quality sCT images from stand-alone thoracic CBCT images in an efficient way and thus has the potential to support CBCT-based esophageal cancer adaptive radiotherapy.

Radiation pneumonitis prediction after stereotactic body radiation therapy based on 3D dose distribution: dosiomics and/or deep learning-based radiomics features.

BACKGROUND: This study was designed to establish radiation pneumonitis (RP) prediction models using dosiomics and/or deep learning-based radiomics (DLR) features based on 3D dose distribution. METHODS: A total of 140 patients with non-small cell lung cancer who received stereotactic body radiation therapy (SBRT) were retrospectively included in this study. These patients were randomly divided into the training (n = 112) and test (n = 28) sets. Besides, 107 dosiomics features were extracted by Pyradiomics, and 1316 DLR features were extracted by ResNet50. Feature visualization was performed based on Spearman's correlation coefficients, and feature selection was performed based on the least absolute shrinkage and selection operator. Three different models were constructed based on random forest, including (1) a dosiomics model (a model constructed based on dosiomics features), (2) a DLR model (a model constructed based on DLR features), and (3) a hybrid model (a model constructed based on dosiomics and DLR features). Subsequently, the performance of these three models was compared with receiver operating characteristic curves. Finally, these dosiomics and DLR features were analyzed with Spearman's correlation coefficients. RESULTS: In the training set, the area under the curve (AUC) of the dosiomics, DLR, and hybrid models was 0.9986, 0.9992, and 0.9993, respectively; the accuracy of these three models was 0.9643, 0.9464, and 0.9642, respectively. In the test set, the AUC of these three models was 0.8462, 0.8750, and 0.9000, respectively; the accuracy of these three models was 0.8214, 0.7857, and 0.8571, respectively. The hybrid model based on dosiomics and DLR features outperformed other two models. Correlation analysis between dosiomics features and DLR features showed weak correlations. The dosiomics features that correlated DLR features with the Spearman's rho |ρ| ≥ 0.8 were all first-order features. CONCLUSION: The hybrid features based on dosiomics and DLR features from 3D dose distribution could improve the performance of RP prediction after SBRT.

Voxel-Level BED Corrected Dosimetric and Radiobiological Assessment of 2 Kinds of Hybrid Radiotherapy Planning Methods for Stage III NSCLC.

Background/purpose: To access the comparative dosimetric and radiobiological advantages of two methods of intensity-modulated radiation therapy (IMRT)-based hybrid radiotherapy planning for stage III nonsmall cell lung cancer (NSCLC). Methods: Two hybrid planning methods were respectively characterized by conventional fraction radiotherapy (CFRT) and stereotactic body radiotherapy (SBRT) and CFRT and simultaneous integrated boost (SIB) planning. All plans were retrospectively completed using the 2 methods for 20 patients with stage III NSCLC. CFRT and SBRT dose regimes 2 Gy × 30 f and 12.5 Gy × 4 f were, respectively, used for planning target volume of lymph node (PTVLN) and planning target volume of the primary tumor (PTVPT), while dose regimes 2 Gy × 26 f for PTVLN and sequential 2 Gy × 4 f for PTVLN combined with 12.5 Gy × 4 f for PTVPT were adopted for CFRT and SIB plans. SBRT and SIB EQD2 dose were calculated voxel by voxel, and then, respectively, superimposed with 30-fraction and 26-fraction CFRT plan dose to achieve biological equivalent dose (BED) dosimetric parameters of CFRT and SBRT and CFRT and SIB plans. Tumor control probability (TCP)/normal tissue complication probability (NTCP) was, respectively, calculated by equivalent uniform dose/Lyman-Kutcher-Burman models. BED plan parameters and TCP/NTCP were analyzed between 2 methods of hybrid planning. Primary tumor/lymph node (LN)/total TCP values were, respectively, evaluated as a function of the radiation dose needed to control 50% of tumor (TCD50) for 20 patients. Dosimetric errors were analyzed by nontransit electronic portal imaging device dosimetry measurement during hybrid plan delivery. Results: Statistically lower BED plan parameters of PTVLN D2 and homogeneity index resulted in slightly lower averaged LN/total TCP curves by CFRT and SIB planning. The gaps between Max and Min LN/total TCP curves were significantly closer for CFRT and SIB planning, which indicated better robustness of LN/total TCPs. A lower esophagus dose resulted in a lower esophagus NTCP by CFRT and SIB planning, which may be compromised by 1 week shorter overall treatment time by CFRT and SIB irradiation. Spinal cord Dmax was significantly reduced by CFRT and SIB plans. The dose verification results of the subplans involved in hybrid plans were acceptable, which showed that the 2 methods of hybrid planning could be delivered accurately in our center. Conclusion: CFRT and SIB plannings have more advantages on BED plan parameters and TCP/NTCP than CFRT and SBRT planning, and both methods of IMRT-based hybrid planning could be executed accurately for stage III NSCLC. The effectiveness of the results needs to be validated in the hybrid trial.

Deep Learning for Patient-Specific Quality Assurance: Predicting Gamma Passing Rates for IMRT Based on Delivery Fluence Informed by log Files.

Objectives: In this study, we propose a deep learning-based approach to predict Intensity-modulated radiation therapy (IMRT) quality assurance (QA) gamma passing rates using delivery fluence informed by log files. Methods: A total of 112 IMRT plans for chest cancers were planned and measured by portal dosimetry equipped on TrueBeam linac. The convolutional neural network (CNN) based learning model was trained using delivery fluence as inputs and gamma passing rates (GPRs) of 4 different criteria (3%/3 mm, 2%/3 mm, 3%/2 mm, and 2%/2 mm) as outputs. Model performance for both validation and test sets was assessed using mean absolute error (MAE), mean squared error (MSE), root MSE (RMSE), Spearman rank correlation coefficients (Sr), and Determination coefficient (R2) between the measured and predicted GPR values. Results: In the test set, the MAE of the prediction model were 0.402, 0.511, 1.724, and 2.530, the MSE were 0.640, 0.986, 6.654, and 9.508, the RMSE were 0.800, 0.993, 2.580, and 3.083, the Sr were 0.643, 0.684, 0.821, and 0.824 (P < .001) and the R2 were 0.4110, 0.4666, 0.6677, and 0.6769 for 3%/3 mm, 3%/2 mm, 2%/3 mm, and 2%/2 mm, respectively. The MAE and RMSE of the prediction model decreased with stricter gamma criteria while the Sr and R2 between measured and predicted GPR values increased. Conclusions: The CNN prediction model based on delivery fluence informed by log files could accurately predict IMRT QA passing rates for different gamma criteria. It could reduce QA workload and improve efficiency in pretreatment QA. Our results suggest that the CNN prediction model based on delivery fluence informed by log files may be a promising tool for the gamma evaluation of IMRT QA.

Nanoparticle radiosensitization: from extended local effect modeling to a survival modification framework of compound Poisson additive killing and its carbon dots validation.

Objective. To construct an analytical model instead of local effect modeling for the prediction of the biological effectiveness of nanoparticle radiosensitization.Approach. An extended local effects model is first proposed with a more comprehensive description of the nanoparticles mediated local killing enhancements, but meanwhile puts forward challenging issues that remain difficult and need to be further studied. As a novel method instead of local effect modeling, a survival modification framework of compound Poisson additive killing is proposed, as the consequence of an independent additive killing by the assumed equivalent uniform doses of individual nanoparticles per cell under the LQ model. A compound Poisson killing (CPK) model based on the framework is thus derived, giving a general expression of nanoparticle mediated LQ parameter modification. For practical use, a simplified form of the model is also derived, as a concentration dependent correction only to theαparameter, with the relative correction (α″/α) dominated by the mean number, and affected by the agglomeration of nanoparticles per cell. For different agglomeration state, a monodispersion model of the dispersity factorη = 1, and an agglomeration model of 2/3 < Î· < 1, are provided for practical prediction of (α″/α) value respectively.Main results. Initial validation by the radiosensitization of HepG2 cells by carbon dots showed a high accuracy of the CPK model. In a safe range of concentration (0.003-0.03µgµl-1) of the carbon dots, the prediction errors of the monodispersion and agglomeration models were both within 2%, relative to the clonogenic survival data of the sensitized HepG2 cells.Significance. The compound Poisson killing model provides a novel approach for analytical prediction of the biological effectiveness of nanoparticle radiosensitization, instead of local effect modeling.

Dosimetric effect of intensity-modulated radiation therapy for postoperative non-small cell lung cancer with and without air cavity in the planning target volume.

To evaluate the dosimetric effect of intensity-modulated radiation therapy (IMRT) for postoperative non-small cell lung cancer (NSCLC), with and without the air cavity in the planning target volume (PTV). Two kinds of IMRT plans were made for 21 postoperative NSCLC patients. In Plan-0: PTV included the tracheal air cavity, and in Plan-1: the air cavity was subtracted from the PTV. For PTV, the dosimetric parameters, including Dmean, D98, D95, D2, D0.2, conformity index (CI), and homogeneity index (HI) were evaluated. For organs at risk (OARs), the evaluation indexes, included the V5, V20 and the mean lung dose (MLD) of total lung, the V30, V40, and the mean heart dose (MHD) of heart, the spinal cord Dmax, and the V35 and the mean esophageal dose (MED) of esophagus. The number of segments and MUs were also recorded. Additionally, the correlation between the Plan-1 dosimetric change value relative to Plan-0, the size of air cavity, and the volume proportion of the cavity in the PTV was also analyzed. The Dmean of PTV, D2, D0.2, HI and CI in Plan-1 decreased compared with those in Plan-0. For OARs, the V30, MHD, and MED also decreased. The CI change value of Plan-1 relative to Plan-0 had a significantly negative correlation with the size and the volume proportion of air cavity, and the MED change value also had a significantly negative correlation with the air cavity size. The IMRT plans for patients with postoperative NSCLC can achieve a better target dose distribution and offer an improved sparing of the heart and esophagus by removing the PTV air cavity, while reducing the target conformity. The change value of CI and MED had a significantly negative correlation with the air cavity size.

How Does the Gradient Measure of the Lung SBRT Treatment Plan Depend on the Tumor Volume and Shape?

PURPOSE: Gradient measure (GM) is a critical index related to normal tissue sparing in radiosurgery. This study aims to describe the dependence of GM on target volume and target shape for lung stereotactic body radiation therapy (SBRT) treatment plans. METHODS: A total of 307 peripheral and 119 central lung SBRT treatment plans were enrolled for this study. A least-squares regression was used for data analysis. First, the equations with different functional forms were established to determine the dependence of GM on a univariaty (VP or Sp) and bivariaty (VP and Sp), respectively. Then, the correlation coefficients and p-values of variables for all equations were compared and analyzed to determine the dependence of GM on PTV volume (VP) and sphericity (Sp). RESULTS: The power equations had the highest coefficient of determination (R2) in the dependence results of GM on univariate VP. The equations were GM = 0.674 V P 0.178 and GM = 0.660 V P 0.185 for peripheral and central lesions, respectively. On the other hand, the R2 of all functional forms were less than 0.25 when the relationship of GM versus univariate Sp was analyzed. Similarly, the power equation also obtained the highest R2 in bivariaty VP and Sp analysis, whether for central or peripheral. However, the R2 of the bivariate equations were not improved compared with those of univariate equations. Moreover, the p-values of the variable Sp were greater than 0.05. CONCLUSIONS: The GM of the lung SBRT plan is shape-independent and volume-dependent. The dependence of GM on PTV volume for peripheral and central lung cancer can be described by two different power equations. The results of this study can be used as a potential tool to assist dosimetric quality control during the radiosurgery process.

Account for the Full Extent of Esophagus Motion in Radiation Therapy Planning: A Preliminary Study of the IRV of the Esophagus.

OBJECTIVE: Accounting for esophagus motion in radiotherapy planning is an important basis for accurate assessment of toxicity. In this study, we calculated how much the delineations of the esophagus should be expanded based on three-dimensional (3D) computed tomography (CT), four-dimensional (4D) average projection (AVG), and maximum intensity projection (MIP) scans to account for the full extent of esophagus motion during 4D imaging acquisition. METHODS AND MATERIALS: The 3D and 4D CT scans of 20 lung cancer patients treated with conventional radiotherapy and 20 patients treated with stereotactic ablative radiation therapy (SBRT) were used. Radiation oncologists contoured the esophagus on the 3DCT, AVG, MIP and 25% exhale scans, and the combination of the esophagus in every phase of 4DCT. The union of all 4D phase delineations (U4D) represented the full extent of esophagus motion during imaging acquisition. Surface distances from U4D to 3D, AVG, and MIP volumes were calculated. Distances in the most extreme surface points (1.5 cm most superoinferior, 10% most right/left/anteroposterior) were used to derive margins accounting only for systematic (delineation) errors. RESULTS: Esophagus delineations on the MIP were the closest to the full extent of motion, requiring only 6.9 mm margins. Delineations on the AVG and 3D scans required margins up to 7.97 and 7.90 mm, respectively. The largest margins were for the inferior, right, and anterior aspects for the delineations on the 3D, AVG, and MIP scans, respectively. CONCLUSION: Delineations on 3D, AVG, or MIP scans required extensions for representing the esophagus's full extent of motion, with the MIP requiring the smallest margins. Research including daily imaging to determine the random components for the margins and dosimetric measurements to determine the relevance of creating a planning organ at risk volume (PRV) of the esophagus is required.

Influence of Clinical and Tumor Factors on Interfraction Setup Errors With Rotation Correction for Vacuum Cushion in Lung Stereotactic Body Radiation Therapy.

PURPOSE: To explore the influence of clinical and tumor factors over interfraction setup errors with rotation correction for non-small cell lung cancer (NSCLC) stereotactic body radiation therapy (SBRT) patients immobilized in vacuum cushion (VC) to better understand whether patient re-setup could further be optimized with these parameters. MATERIALS AND METHODS: This retrospective study was conducted on 142 NSCLC patients treated with SBRT between November 2017 to July 2019 in the local institute. Translation and rotation setup errors were analyzed in 732 cone-beam computed tomography (CBCT) scans before treatment. Differences between groups were analyzed using independent sample t-test. Logistic regression test was used to analyze possible correlations between patient re-setup and clinical and tumor factors. RESULTS: Mean setup errors were the largest in anterior-posterior (AP) direction (3.2 ± 2.4 mm) compared with superior-inferior (SI) (2.8 ± 2.1 mm) and left-right (LR) (2.5 ± 2.0 mm) directions. The mean values were similar in pitch, roll, and rtn directions. Of the fractions, 83.7%, 90.3%, and 86.6% satisfied setup error tolerance limits in AP, SI, and LR directions, whereas 95% had rotation setup errors of <2° in the pitch, roll, or rtn directions. Setup errors were significantly different in the LR direction when age, body mass index (BMI), and "right vs. left" location parameters were divided into groups. Both univariate and multivariable model analyses showed that age (p = 0.006) and BMI (p = 0.002) were associated with patient re-setup. CONCLUSIONS: Age and BMI, as clinical factors, significantly influenced patient re-setup in the current study, whereas all other clinical and tumor factors were not correlated with patient re-setup. The current study recommends that more attention be paid to setup for elderly patients and patients with larger BMI when immobilized using VC, especially in the left-right direction.

Investigation of Predictors to Achieve Acceptable Lung Dose in T-Shaped Upper and Middle Esophageal Cancer With IMRT and VMAT.

PURPOSE: The purpose of this study is to investigate whether there are predictors and cutoff points that can predict the acceptable lung dose using intensity-modulated radiation therapy (IMRT) and volume-modulated arc therapy (VMAT) in radiotherapy for upper ang middle esophageal cancer. MATERIAL AND METHODS: Eighty-two patients with T-shaped upper-middle esophageal cancer (UMEC) were enrolled in this retrospective study. Jaw-tracking IMRT plan (JT-IMRT), full-arc VMAT plan (F-VMAT), and pactial-arc VMAT plan (P-VMAT) were generated for each patient. Dosimetric parameters such as MLD and V20 of total lung were compared among the three plannings. Ten factors such as PCTVinferior length and PCTVinferior length/total lung length were calculated to find the predictors and cutoff points of the predictors. All patients were divided into two groups according to the cutoff points, and the dosimetric differences between the two groups of the three plans were compared. ANOVA, receiver operating characteristic (ROC) analysis, and Mann-Whitney U-test were performed for comparisons between datasets. A p <0.05 was considered statistically significant. RESULT: The quality of the targets of the three plannings was comparable. The total lung dose in P-VMAT was significantly lower than that in JT IMRT and F-VMAT. Monitor unit (MU) of F-VMAT and P-VMAT was significantly lower than that of JT IMRT. ROC analysis showed that among JT IMRT, F-VMAT, and P-VMAT, PCTVi-L, and PCTVi-L/TLL had diagnostic power to predict the suitability of RT plans according to lung dose constraints of our department. For JT IMRT, the cutoff points of PCTVi-L and PCTVi-L/TLL were 16.6 and 0.59. For F-VMAT, the cutoff points of PCTVi-L and PCTVi-L/TLL were 16.75 and 0.62. For P-VMAT, the cutoff points of PCTVi-L and PCTVi-L/TLL were 15.15 and 0.59. After Mann-Whitney U-test analysis, it was found that among the three plannings, the group with lower PCTVi-L and PCTVi-L/TLL could significantly reduce the dose of total lung and heart (p <0.05). CONCLUSION: PCTVi-L <16.6 and PCTVi-L/TLL <0.59 for JT IMRT, PCTVi-L <16.75 and PCTVi-L/TLL <0.62 for F-VMAT and PCTVi-L <15.15, and PCTVi-L/TLL <0.59 for P-VMAT can predict whether patients with T-shaped UMEC can meet the lung dose limits of our department.

A novel CRT-IMRT-combined (Co-CRIM) planning technique for peripheral lung stereotactic body radiotherapy in pinnacle treatment planning system.

OBJECTIVES: This study attempts to explore a novel peripheral lung stereotactic body radiotherapy (SBRT) planning technique that can balance the pros and cons of three-dimensional conformal radiotherapy (CRT) and intensity-modulated radiation therapy (IMRT) / volumetric modulated arc therapy (VMAT). METHODS: Treatment plans were retrospectively designed based on CRT, IMRT, VMAT, and the proposed CRT-IMRT-combined (Co-CRIM) techniques using Pinnacle treatment planning system (TPS) for 20 peripheral lung cancer patients. Co-CRIM used an inverse optimization algorithm available in Pinnacle TPS. To develop a Co-CRIM plan, the number of segments in each field was limited to one, the minimum segment area was set to the internal target volume (ITV), and the minimum monitor units (MU) of the segment was the quotient of fractional dose divided by twice the number of total fields. The performance of Co-CRIM was then compared with other techniques. RESULTS: For conformity index (CI), Co-CRIM performed comparably to IMRT/VMAT but better than CRT. For gradient index (GI), Co-CRIM was similar to IMRT/VMAT or CRT. For heterogeneity index (HI), Co-CRIM was comparable to IMRT/VMAT, higher than CRT. The dosimetric results of spinal cord and lung with Co-CRIM were better than CRT, comparable to IMRT, but inferior to VMAT. The MU resulted from Co-CRIM was lower than IMRT/VMAT but higher than CRT. For plan verification γ passing rate, Co-CRIM was higher than IMRT/VMAT, comparable to CRT. For planning time, Co-CRIM was shorter than CRT or VMAT but similar to IMRT. CONCLUSIONS: The proposed Co-CRIM technique on Pinnacle TPS is an effective planning technique for peripheral lung SBRT.

Isotoxic investigation of 18F-FDG PET/CT-guided dose escalation with intensity-modulated radiotherapy for LA-NSCLC.

PURPOSE: This research compared differences of dosimetric and biological parameters between PET/CT-guided isotoxic SIB-IMRT plans and conventional radiotherapy plans for patients with LA-NSCLC, and it also evaluated the factors that affect dose escalation. MATERIALS AND METHODS: This study consisted of a retrospective cohort of thirty patients with IIIA-IIIB NSCLC. SIB-IMRT (Plan_iso) and conventional radiotherapy (Plan_primary) plans were generated using auto-planning. Dosimetric parameters such as mean lung dose (MLD) and other indicators were compared. Tumor control probability (TCP) of PTV and normal tissue complication probability (NTCP) of total lung, heart, esophagus, and spinal cord were calculated. The relationships between dose escalation and 3 D length of PTV and other factors were analyzed. Paired-samples t-test, Mann-Whitney U test, and Chi-Square test were performed for comparisons between datasets. A P < .05 was considered statistically significant. RESULTS: The dosimetric parameters of PTV in Plan_iso were higher than those of PTV in Plan_primary, and there were significant differences (p < .05). Compared with Plan_primary, Plan_iso slightly increased dosimetric parameters of the total lung, heart, spinal cord, esophagus, and MUs. The absolute differences were small. TCPs of PTV in Plan_iso were significantly higher than those in Plan_primary. NTCPs of the total lung, esophagus, and spinal cord in Plan_iso were higher than those in Plan_primary. There were significant differences, but the absolute differences were small. NTCP of heart in Plan_iso was slightly higher than that in Plan_primary, but there was no statistical difference. CONCLUSIONS: For LA-NSCLC, the SIB based on isotoxic radiotherapy can significantly increase TCP under the premise that the toxicity of OARs is comparable.

A 2D-3D hybrid convolutional neural network for lung lobe auto-segmentation on standard slice thickness computed tomography of patients receiving radiotherapy.

BACKGROUND: Accurate segmentation of lung lobe on routine computed tomography (CT) images of locally advanced stage lung cancer patients undergoing radiotherapy can help radiation oncologists to implement lobar-level treatment planning, dose assessment and efficacy prediction. We aim to establish a novel 2D-3D hybrid convolutional neural network (CNN) to provide reliable lung lobe auto-segmentation results in the clinical setting. METHODS: We retrospectively collected and evaluated thorax CT scans of 105 locally advanced non-small-cell lung cancer (NSCLC) patients treated at our institution from June 2019 to August 2020. The CT images were acquired with 5 mm slice thickness. Two CNNs were used for lung lobe segmentation, a 3D CNN for extracting 3D contextual information and a 2D CNN for extracting texture information. Contouring quality was evaluated using six quantitative metrics and visual evaluation was performed to assess the clinical acceptability. RESULTS: For the 35 cases in the test group, Dice Similarity Coefficient (DSC) of all lung lobes contours exceeded 0.75, which met the pass criteria of the segmentation result. Our model achieved high performances with DSC as high as 0.9579, 0.9479, 0.9507, 0.9484, and 0.9003 for left upper lobe (LUL), left lower lobe (LLL), right upper lobe (RUL), right lower lobe (RLL), and right middle lobe (RML), respectively. The proposed model resulted in accuracy, sensitivity, and specificity of 99.57, 98.23, 99.65 for LUL; 99.6, 96.14, 99.76 for LLL; 99.67, 96.13, 99.81 for RUL; 99.72, 92.38, 99.83 for RML; 99.58, 96.03, 99.78 for RLL, respectively. Clinician's visual assessment showed that 164/175 lobe contours met the requirements for clinical use, only 11 contours need manual correction. CONCLUSIONS: Our 2D-3D hybrid CNN model achieved accurate automatic segmentation of lung lobes on conventional slice-thickness CT of locally advanced lung cancer patients, and has good clinical practicability.