Dose-volume Histogram Indices Research Articles

Dose Evaluation in 2-Phase Method for Advanced Esophageal Cancer by Hybrid Irradiation Techniques

PurposeIn concurrent chemoradiotherapy for advanced esophageal cancer, a 2-phase method consisting of initial irradiation of a wide elective nodal region and boost irradiation of the primary lesion is commonly employed. Although dose escalation to the primary lesion may be required to achieve higher local control rates, the radiation dose to critical organs must not exceed dose constraints. To achieve an optimum balance of dose prescription and dose reduction to surrounding organs, such as the lungs and heart, we compared hybrid dose distributions and investigated the best combination of the following recent irradiation techniques: volumetric modulation arc therapy (VMAT), proton broad-beam irradiation, and intensity-modulated proton beam therapy (IMPT). Materials and MethodsForty-five patients with advanced esophageal cancer whose primary lesions were located in the middle- or lower-thoracic region were studied. Radiotherapy plans for the initial and boost irradiation in the 2-phase method were calculated using VMAT, proton broad-beam irradiation, and IMPT calculation codes, and the dose-volume histogram indices of the lungs and heart for the accumulated plans were compared. ResultsIn plans using boost proton irradiation with a prescribed dose of 60 Gy(RBE), all dose-volume histogram indices were significantly below the tolerance limits. Initial and boost irradiation with VMAT resulted in the median dose of V30 Gy(RBE)(heart) of 27.4% and an achievement rate below the tolerance limit of 57.8% (26 cases). In simulations of dose escalation up to 70 Gy(RBE), initial and boost IMPT resulted in the highest achievement rate, satisfying all dose constraints in 95.6% (43 cases). ConclusionApplying VMAT to both initial and boost irradiation is not recommended because of the increased risk of the cardiac dose exceeding the tolerance limit. IMPT may allow dose escalation of up to 70 Gy(RBE) without radiation risks to the lungs and heart in the treatment of advanced esophageal cancer.

Improving Dose Volume Histogram (DVH) Based Analysis of Clinical Outcomes Using Modern Statistical Techniques: A Systematic Answer to Multiple Comparisons Concerns

DVH constraints are essential in the clinical practice of radiation therapy. Historically, DVH constraints were found through sparse sampling of all possible DVH indices to find one that appeared to be most predictive for clinical toxicity. This approach can lead to inconsistent results among studies and to multiple comparison concerns. We aim to solve both problems by examining a full array of DVH indices using statistical methods that account for strong correlations among DVH indices and incorporate radiobiological knowledge constraints. We extracted a dense array of V%_D indices from a treatment planning system using ESAPI interface, with V%_D corresponding to the volume fraction irradiated to dose D, or higher. We used Fused Lasso as the base model to compensate for correlations among DVH indices because it applies a penalty on the difference between DVH variables with adjacent dose. The base model was augmented with additional constraints based on radiobiological considerations: the positivity constraint (beta_i > 0) which assumes that any tissue irradiation cannot reduce the risk of toxicity, and monotonicity constraint (beta_i+1 > = beta_i) which assumes that higher dose to a fixed volume fraction cannot be associated with a lower risk of toxicity. We called the hybrid model KC-Lasso (Knowledge Constrained Lasso) and applied it to two clinical examples: grade 2 acute rectal toxicity in conventionally fractionated RT for 79 prostate cancer patients (77.4 Gy + MR based boost to 81-83 Gy) and cardiac toxicity in conventionally fractionated RT for 119 locally advanced Non-small Cell Lung Cancer (NSCLC) patients (Median prescribed dose 62 Gy). We further examined alternative data driven models to determine the importance of knowledge constraints. KC-Lasso detected two distinct dose thresholds for grade 2 rectal toxicity, at 35 Gy and 78 Gy. A threshold of 51 Gy was detected for reduced overall survival due to cardiac irradiation in NSCLC patients. An examination of KC-Lasso models at varying step size suggested that a single mid-range index can be used as a treatment planning constraint while full model can be used for confirmatory, final plan evaluation. Alternative models which lack knowledge constraints show patterns of negative and isolated coefficients which are difficult to interpret and are not likely to be generalizable. A more systematic approach to the analysis of correlations between DVH constraints and clinical toxicity can lead to greater consistency of results among different studies, better understanding of true dose thresholds and results which are more generalizable.

Beam mask and sliding window-facilitated deep learning-based accurate and efficient dose prediction for pencil beam scanning proton therapy.

Accurate and efficient dose calculation is essential for on-line adaptive planning in proton therapy. Deep learning (DL) has shown promising dose prediction results in photon therapy. However, there is a scarcity of DL-based dose prediction methods specifically designed for proton therapy. Successful dose prediction method for proton therapy should account for more challenging dose prediction problems in pencil beam scanning proton therapy (PBSPT) due to its sensitivity to heterogeneities. To develop a DL-based PBSPT dose prediction workflow with high accuracy and balanced complexity to support on-line adaptive proton therapy clinical decision and subsequent replanning. PBSPT plans of 103 prostate cancer patients (93 for training and the other 10 for independent testing) and 83 lung cancer patients (73 for training and the other 10 for independent testing) previously treated at our institution were included in the study, each with computed tomography scans (CTs), structure sets, and plan doses calculated by the in-house developed Monte-Carlo dose engine (considered as the ground truth in the model training and testing). For the ablation study, we designed three experiments corresponding to the following three methods: (1) Experiment 1, the conventional region of interest (ROI) (composed of targets and organs-at-risk [OARs]) method. (2) Experiment 2, the beam mask (generated by raytracing of proton beams) method to improve proton dose prediction. (3) Experiment 3, the sliding window method for the model to focus on local details to further improve proton dose prediction. A fully connected 3D-Unet was adopted as the backbone. Dose volume histogram (DVH) indices, 3D Gamma passing rates with a criterion of 3%/3mm/10%, and dice coefficients for the structures enclosed by the iso-dose lines between the predicted and the ground truth doses were used as the evaluation metrics. The calculation time for each proton dose prediction was recorded to evaluate the method's efficiency. Compared to the conventional ROI method, the beam mask method improved the agreement of DVH indices for both targets and OARs and the sliding window method further improved the agreement of the DVH indices (for lung cancer, CTV D98 absolute deviation: 0.74 ± 0.18 vs. 0.57 ± 0.21 vs. 0.54 ± 0.15 Gy[RBE], ROI vs. beam mask vs. sliding window methods, respectively). For the 3D Gamma passing rates in the target, OARs, and BODY (outside target and OARs), the beam mask method improved the passing rates in these regions and the sliding window method further improved them (for prostate cancer, targets: 96.93% ± 0.53% vs. 98.88% ± 0.49% vs. 99.97% ± 0.07%, BODY: 86.88% ± 0.74% vs. 93.21% ± 0.56% vs. 95.17% ± 0.59%). A similar trend was also observed for the dice coefficients. This trend was especially remarkable for relatively low prescription isodose lines (for lung cancer, 10% isodose line dice: 0.871 ± 0.027 vs. 0.911 ± 0.023 vs. 0.927 ± 0.017). The dose predictions for all the testing cases were completed within 0.25s. An accurate and efficient deep learning-augmented proton dose prediction framework has been developed for PBSPT, which can predict accurate dose distributions not only inside but also outside ROI efficiently. The framework can potentially further reduce the initial planning and adaptive replanning workload in PBSPT.

Dosimetric Validation for Prospective Clinical Trial of GRID Collimator-Based Spatially Fractionated Radiation Therapy: Dose Metrics Consistency and Heterogeneous Pattern Reproducibility

Dose heterogeneity within a tumor target is likely responsible for the biologic effects and local tumor control from spatially fractionated radiation therapy (SFRT). This study used a commercially available GRID-pattern dose mudulated nonuniform radiation therapy (GRID) collimator to assess the interplan variability of heterogeneity dose metrics in patients with various bulky tumor sizes and depths. The 3-dimensional heterogeneity metrics of 14 bulky tumors, ranging from 155 to 2161 cm3 in volume, 6 to 23 cm in equivalent diameter, and 3 to 13 cm in depth, and treated with GRID collimator-based SFRT were studied. A prescription dose of 15 Gy was given at the tumor center with 6 MV photons. The dose-volume histogram indices, dose heterogeneity parameters, and peak/valley dose ratios were derived; the equivalent uniform doses of cancer cells with various radiosensitivities in each plan were estimated. To account for the spatial fractionation, high dose core number density of the tumor target was defined and calculated. Among 14 plans, the dose-volume histogram indices D5, D10, D50, D90, and D95 (doses covering 5%, 10%, 50%, 90%, and 95% of the target volume) were found within 10% variation. The dose ratio of D10/D90 also showed a moderate consistency (range, 3.9-5.0; mean, 4.4). The equivalent uniform doses were consistent, ranging from 4.3 to 5.5 Gy, mean 4.6 Gy, for radiosensitive cancer cells and from 5.8 to 6.9 Gy, mean 6.2 Gy, for radioresistant cancer cells. The high dose core number density was within 20% among all plans. GRID collimator-based SFRT delivers a consistent heterogeneity dose distribution and high dose core density across bulky tumor plans. The interplan reproducibility and simplicity of GRID therapy may be useful for certain clinical indications and interinstitutional clinical trial design, and its heterogeneity metrics may help guide multileaf-collimator-based SFRT planning to achieve similar or further optimized dose distributions.

Validation of robust radiobiological optimization algorithms based on the mixed beam model for intensity-modulated carbon-ion therapy.

Currently, treatment planning systems (TPSs) that can compute the intensities of intensity-modulated carbon-ion therapy (IMCT) using scanned carbon-ion beams are limited. In the present study, the computational efficacy of the newly designed IMCT algorithms was analyzed for the first time based on the mixed beam model with respect to the physical and biological doses; moreover, the validity and effectiveness of the robust radiobiological optimization were verified. A dose calculation engine was independently generated to validate a clinical dose determined in the TPS. A biological assay was performed using the HSGc-C5 cell line to validate the calculated surviving fraction (SF). Both spot control (SC) and voxel-wise worst-case scenario (WC) algorithms were employed for robust radiobiological optimization followed by their application in a Radiation Therapy Oncology Group benchmark phantom under homogeneous and heterogeneous conditions and a clinical case for range and position errors. Importantly, for the first time, both SC and WC algorithms were implemented in the integrated TPS platform that can compute the intensities of IMCT using scanned carbon-ion beams for robust radiobiological optimization. For assessing the robustness, the difference between the maximum and minimum values of a dose-volume histogram index in the examined error scenarios was considered as a robustness index. The relative biological effectiveness (RBE) determined by the independent dose calculation engine exhibited a -0.6% difference compared with the RBE defined by the TPS at the isocenter, whereas the measured and the calculated SF were similar. Regardless of the objects, compared with the conventional IMCT, the robust radiobiological optimization enhanced the sensitivity of the examined error scenarios by up to 19% for the robustness index. The computational efficacy of the novel IMCT algorithms was verified according to the mixed beam model with respect to the physical and biological doses. The robust radiobiological optimizations lowered the impact of range and position uncertainties considerably in the examined scenarios. The robustness of the WC algorithm was more enhanced compared with that of the SC algorithm. Nevertheless, the SC algorithm can be used as an alternative to the WC IMCT algorithm with respect to the computational cost.

Deep-learning based fast and accurate 3D CT deformable image registration in lung cancer.

Deformable Image Registration (DIR) is an essential technique required in many applications of radiation oncology. However, conventional DIR approaches typically take several minutes to register one pair of 3D CT images and the resulting deformable vector fields (DVFs) are only specific to the pair of images used, making it less appealing for clinical application. A deep-learning-based DIR method using CT images is proposed for lung cancer patients to address the common drawbacks of the conventional DIR approaches and in turn can accelerate the speed of related applications, such as contour propagation, dose deformation, adaptive radiotherapy (ART), etc. METHODS: A deep neural network based on VoxelMorph was developed to generate DVFs using CT images collected from 114 lung cancer patients. Two models were trained with the weighted mean absolute error (wMAE) loss and structural similarity index matrix (SSIM) loss (optional) (i.e., the MAE model and the M+S model). In total, 192 pairs of initial CT (iCT) and verification CT (vCT) were included as a training dataset and the other independent 10 pairs of CTs were included as a testing dataset. The vCTs usually were taken 2weeks after the iCTs. The synthetic CTs (sCTs) were generated by warping the vCTs according to the DVFs generated by the pre-trained model. The image quality of the synthetic CTs was evaluated by measuring the similarity between the iCTs and the sCTs generated by the proposed methods and the conventional DIR approaches, respectively. Per-voxel absolute CT-number-difference volume histogram (CDVH) and MAE were used as the evaluation metrics. The time to generate the sCTs was also recorded and compared quantitatively. Contours were propagated using the derived DVFs and evaluated with SSIM. Forward dose calculations were done on the sCTs and the corresponding iCTs. Dose volume histograms (DVHs) were generated based on dose distributions on both iCTs and sCTs generated by two models, respectively. The clinically relevant DVH indices were derived for comparison. The resulted dose distributions were also compared using 3D Gamma analysis with thresholds of 3mm/3%/10% and 2mm/2%/10%, respectively. The two models (wMAE and M+S) achieved a speed of 263.7±163 / 265.8±190ms and a MAE of 13.15±3.8 / 17.52±5.8 HU for the testing dataset, respectively. The average SSIM scores of 0.987±0.006 and 0.988±0.004 were achieved by the two proposed models, respectively. For both models, CDVH of a typical patient showed that less than 5% of the voxels had a per-voxel absolute CT-number-difference larger than 55 HU. The dose distribution calculated based on a typical sCT showed differences of ≤2cGy[RBE] for clinical target volume (CTV) D95 and D5 , within ±0.06% for total lung V5 , ≤1.5cGy[RBE] for heart and esophagus Dmean , and ≤6cGy[RBE] for cord Dmax compared to the dose distribution calculated based on the iCT. The good average 3D Gamma passing rates (>96% for 3mm/3%/10% and>94% for 2mm/2%/10%, respectively) were also observed. A deep neural network-based DIR approach was proposed and has been shown to be reasonably accurate and efficient to register the initial CTs and verification CTs in lung cancer.

A generalization performance study on the boosting radiotherapy dose calculation engine based on super-resolution

PurposeDuring the radiation treatment planning process, one of the time-consuming procedures is the final high-resolution dose calculation, which obstacles the wide application of the emerging online adaptive radiotherapy techniques (OLART). There is an urgent desire for highly accurate and efficient dose calculation methods. This study aims to develop a dose super resolution-based deep learning model for fast and accurate dose prediction in clinical practice. MethodA Multi-stage Dose Super-Resolution Network (MDSR Net) architecture with sparse masks module and multi-stage progressive dose distribution restoration method were developed to predict high-resolution dose distribution using low-resolution data. A total of 340 VMAT plans from different disease sites were used, among which 240 randomly selected nasopharyngeal, lung, and cervix cases were used for model training, and the remaining 60 cases from the same sites for model benchmark testing, and additional 40 cases from the unseen site (breast and rectum) was used for model generalizability evaluation. The clinical calculated dose with a grid size of 2 mm was used as baseline dose distribution. The input included the dose distribution with 4 mm grid size and CT images. The model performance was compared with HD U-Net and cubic interpolation methods using Dose-volume histograms (DVH) metrics and global gamma analysis with 1%/1 mm and 10% low dose threshold. The correlation between the prediction error and the dose, dose gradient, and CT values was also evaluated. ResultsThe prediction errors of MDSR were 0.06–0.84% of Dmean indices, and the gamma passing rate was 83.1–91.0% on the benchmark testing dataset, and 0.02–1.03% and 71.3–90.3% for the generalization dataset respectively. The model performance was significantly higher than the HD U-Net and interpolation methods (p < 0.05). The mean errors of the MDSR model decreased (monotonously by 0.03–0.004%) with dose and increased (by 0.01–0.73%) with the dose gradient. There was no correlation between prediction errors and the CT values. ConclusionThe proposed MDSR model achieved good agreement with the baseline high-resolution dose distribution, with small prediction errors for DVH indices and high gamma passing rate for both seen and unseen sites, indicating a robust and generalizable dose prediction model. The model can provide fast and accurate high-resolution dose distribution for clinical dose calculation, particularly for the routine practice of OLART.

Feasibility of online adaptive HDR prostate brachytherapy: A novel treatment concept

PurposeThe purpose of this study was to determine the feasibility of online adaptive transrectal ultrasound (TRUS)-based high-dose-rate prostate brachytherapy (HDRPBT) through retrospective simulation of source positioning and catheter swap errors on patient treatment plans. MethodSource positioning errors (catheter shifts in 1 mm increments in the cranial/caudal, anterior/posterior, and medial/lateral directions up to ±6 mm) and catheter swap errors (between the most and least heavily weighted) were introduced retrospectively into DICOM treatment plans of 20 patients that previously received TRUS HDRPBT. Dose volume histogram (DVH) indices were monitored as errors were introduced sequentially into individual catheters, simulating potential errors throughout treatment. Whenever DVH indices were outside institution thresholds: prostate V100% <95%, urethra D0.1cc >118% and rectum Dmax >80%, the plan was adapted using remaining catheters (i.e., simulating previous catheters as previously delivered). The final DVH indices were recorded. ResultsProstate coverage (V100% >95%) could be maintained for source position errors up to 6 mm through online plan adaptation. The source position error at which the urethra D0.1cc and rectum Dmax was able to return to clinically acceptable levels using online adaptation varied between 6 mm to 1 mm, depending on the direction of the source position error and patient anatomy. After introduction of catheter swap errors to patient plans, prostate V100% was recoverable using online adaptation to near original plan characteristics. Urethra D0.1cc and rectum Dmax showed less recoverability. ConclusionOnline adaptive HDRPBT maintains the prostate V100% to clinically acceptable values for majority of directional shifts. However, the current online adaptive method may not correct for source position errors near organs at risk.

HDR prostate brachytherapy plan robustness and its effect on in-vivo source tracking error thresholds: A multi-institutional study.

PurposeThe purpose of this study was to examine the effect of departmental planning techniques on appropriate in‐vivo source tracking error thresholds for high dose rate (HDR) prostate brachytherapy (BT) treatments, and to determine if a single in‐vivo source tracking error threshold would be appropriate for the same patient anatomy.MethodsThe prostate, rectum, and urethra were contoured on a single patient transrectal ultrasound (TRUS) dataset. Anonymized DICOM files were disseminated to 16 departments who created an HDR prostate BT treatment plan on the dataset with a prescription dose of 15 Gy in a single fraction. Departments were asked to follow their own local treatment planning guidelines. Source positioning errors were then simulated in the 16 treatment plans and the effect on dose–volume histogram (DVH) indices calculated. Change in DVH indices were used to determine appropriate in‐vivo source tracking error thresholds. Plans were considered to require intervention if the following DVH conditions occurred: prostate V100% < 90%, urethra D0.1cc > 118%, and rectumtt Dmax > 80%.ResultsThere was wide variation in appropriate in‐vivo source tracking error thresholds among the 16 participating departments, ranging from 1 to 6 mm. Appropriate in‐vivo source tracking error thresholds were also found to depend on the direction of the source positioning error and the endpoint. A robustness parameter was derived, and found to correlate with the sensitivity of plans to source positioning errors.ConclusionsA single HDR prostate BT in‐vivo source tracking error threshold cannot be applied across multiple departments, even for the same patient anatomy. The burden on in‐vivo source tracking devices may be eased through improving HDR prostate BT plan robustness during the plan optimisation phase.

3D dose reconstruction based on in vivo dosimetry for determining the dosimetric impact of geometric variations in high-dose-rate prostate brachytherapy

IntroductionIn vivo dosimetry (IVD) can be used for source tracking (ST), i.e., estimating source positions, during brachytherapy. The aim of this study was to exploit IVD-based ST to perform 3D dose reconstruction for high-dose-rate prostate brachytherapy and to evaluate the robustness of the treatments against observed geometric variations. Materials and MethodsTwenty-three fractions of high-dose-rate prostate brachytherapy were analysed. The treatment planning was based on MRI. Time-resolved IVD was performed using a fibre-coupled scintillator. ST was retrospectively performed using the IVD measurements. The ST identified 2D positional shifts of each treatment catheter and thereby inferred updated source positions. For each fraction, the dose was recalculated based on the source-tracked catheter positions and compared with the original plan dose using differences in dose volume histogram indices. ResultsOf 352 treatment catheters, 344 had shifts of less than 5 mm. Shifts between 5 and 10 mm were observed for 3 catheters, and shifts greater than 10 mm for 2 catheters. The ST failed for 3 catheters.The maximum relative difference in clinical target volume (prostate + 3 mm isotropic margin) D90% was 5%. In one fraction, the bladder D2cm3 dose increased by 18% (1.4 Gy) due to a single source position being inside the bladder rather than nearby as planned. The max increase in urethra dose was 1.5 Gy (15%). ConclusionIVD-based 3D dose reconstruction for high-dose-rate prostate brachytherapy is feasible. The dosimetric impact of the observed catheter shifts was limited. Dose reconstruction can therefore aid in determining the dosimetric impact of geometric variations and errors in brachytherapy.

Commissioning of GPU-based multi-criteria optimizer combined with plan navigation tools for high-dose-rate brachytherapy.

PurposeRecently, our GPU-based multi-criteria optimization (gMCO) algorithm has been integrated in a graphical user interface (gMCO-GUI) that allows real-time plan navigation through a gMCO-generated set of Pareto-optimal plans for high-dose-rate (HDR) brachytherapy. This work reports on the commissioning of the gMCO algorithm into clinical workflow.Material and methodsOur MCO workflow was validated against Oncentra Prostate v. 4.2.2 (OcP) and Oncentra Brachy v. 4.6.0 (OcB). 40 HDR prostate brachytherapy patients (20 with OcP and 20 with OcB) were retrospectively re-planned with gMCO algorithm by generating 2,000 Pareto-optimal plans. A single gMCO treatment plan was exported using gMCO-GUI plan navigation tools. The optimized dwell positions and dwell times of gMCO plans were exported via DICOM RTPLAN files to OcP/OcB, where final dosimetry was calculated. TG43 implementation in gMCO was validated against the consensus data of flexisource. Five analytical shapes were used as the ground truth for volume calculations. Dose-volume histogram (DVH) curves generated by gMCO were compared with the ones generated by OcP/OcB. 3D dose distributions (and isodose lines) were validated against OcP/OcB using dice similarity coefficient (DSC), 95% undirected Hausdorff distance (95% HD), and γ analysis.ResultsDifferences between –0.4% and 0.3% were observed between gMCO calculated dose rates and the flexisource consensus data. gMCO volumes were within ±2% agreement in 3/5 volumes (deviations within –2.9% and 0.1%). For 9 key DVH indices, the differences between gMCO and OcP/OcB were within ±1.2%. Regarding the accuracy of key isodose lines, the mean DSC was greater than 0.98, and the mean 95% HD was below 0.4 mm. The fraction of voxels with γ ≤ 1 was greater than 99% for all cases with 1%/1 mm threshold.ConclusionsThe GPU-based MCO workflow was successfully integrated into the clinical workflow and validated against OcP and OcB.

Per-voxel constraints to minimize hot spots in linear energy transfer-guided robust optimization for base of skull head and neck cancer patients in IMPT.

Due to the employment of quadratic programming using soft constraints to implement dose volume constraints and the "trial-and-error" procedure needed to achieve a clinically acceptable plan, conventional dose volume constraints (upper limit) are not adequately effective in controlling small and isolated hot spots in the dose/linear energy transfer (LET) distribution. Such hot spots can lead to adverse events. In order to mitigate the risk of brain necrosis, one of the most clinically significant adverse events in patients receiving intensity-modulated proton therapy (IMPT) for base of skull (BOS) cancer, we propose per-voxel constraints to minimize hot spots in LET-guided robust optimization. Ten BOS cancer patients treated with IMPT were carefully selected by meeting one of the following conditions: (1) diagnosis of brain necrosis during follow-up; and (2) considered high risk for brain necrosis by not meeting dose constraints to the brain. An optimizing structure (BrainOPT) and an evaluating structure (BrainROI) that both contained the aforementioned hot dose regions in the brain were generated for optimization and evaluation, respectively. Two plans were generated for every patient: one using conventional dose-only robust optimization, the other using LET-guided robust optimization. The impact of LET was integrated into the optimization via a term of extra biological dose (xBD). A novel optimization tool of per-voxel constraints to control small and isolated hot spots in either the dose, LET, or combined (dose/LET) distribution was developed and used to minimize dose/LET hot spots of the selected structures. Indices from dose-volume histogram (DVH) and xBD dose-volume histogram (xBDVH) were used in the plan evaluation. A newly developed tool of the dose-LET-volume histogram (DLVH) was also adopted to illustrate the underlying mechanism. Wilcoxon signed-rank test was used for statistical comparison of the DVH and xBDVH indices between the conventional dose-only and the LET-guided robustly optimized plans. Per-voxel constraints effectively and efficiently minimized dose hot spots in both dose-only and LET-guided robust optimization and LET hot spots in LET-guided robust optimization. Compared to the conventional dose-only robust optimization, the LET-guided robust optimization could generate plans with statistically lower xBD hot spots in BrainROI (VxBD,50Gy[RBE], p=0.009; VxBD,60Gy[RBE], p=0.025; xBD1cc, p=0.017; xBD2cc, p=0.022) with comparable dose coverage, dose hot spots in the target, and dose hot spots in BrainROI. DLVH analysis indicated that LET-guided robust optimization could either reduce LET at the same dose level or redistribute high LET from high dose regions to low dose regions. Per-voxel constraint is a powerful tool to minimize dose/LET hot spots in IMPT. The LET-guided robustly optimized plans outperformed the conventional dose-only robustly optimized plans in terms of xBD hot spots control.

Exploratory Investigation of Dose-Linear Energy Transfer (LET) Volume Histogram (DLVH) for Adverse Events Study in Intensity-Modulated Proton Therapy (IMPT)

We proposed a novel tool of dose-linear-energy-transfer (LET) volume histogram (DLVH) and performed an exploratory study to investigate rectal bleeding in prostate cancer treated by intensity-modulated proton therapy (IMPT).DLVH was constructed with dose and LET as two axes, while the normalized volume of the structure was contoured in the dose-LET plane as iso-volume lines. We defined DLVH index, DLv% (d,l), i.e., v% of the structure have a dose of d ≥ Gy and an LET of l ≥ keV/µm similar to dose-volume histogram index Dv%. Nine prostate cancer patients with rectal bleeding (CTCAE grade ≥2) were included as the adverse event group, while 48 patients with no complication were considered as the control group. P-value map was constructed by comparison of the DLVH indices of all patients between the two groups using the Mann-Whitney U test. Dose-LET volume constraints (DLVCs) were derived based on the P-value map with a manual selection procedure facilitated by Spearman's correlation tests. The obtained DLVCs were further cross-validated using a multivariate support-vector-machine (SVM)-based normal tissue complication probability (NTCP) model with an independent testing dataset composed of 8 AE and 13 control patients.We extracted two DLVC constraints. One DLVC was obtained: V (dose/LET boundary: 2.5keV/μm at 75 Gy to 3.2keV/μm at 8.65 Gy) < 1.27% (DLVC1), revealing high LET volume effect. The second DLVC was observed: V(72.2 Gy,0keV/μm) < 2.23% (DVLC2), revealing high dose volume effect. The SVM-based NTCP model with two DLVCs provided slightly superior performance on training, validation, and testing compared to the model using generalized equivalent uniform dose (gEUD) alone (Table 1). The corresponding area-under-the-curves (AUCs) of the receiver operating characteristic (ROC) curve were 0.785 vs. 0.762, 0.720 vs. 0.704, and 0.798 vs. 0.779 for the training, validation, and testing dataset, respectively. Sensitivity of 0.67 and specificity of 0.69 were achieved at the NTCP cutoff value of 0.5 in validation.Our results demonstrated the importance of rectal "hot spots" in both high LET (DLVC1) and high dose (DLVC2) in inducing rectal bleeding. The SVM-based NTCP model confirmed the derived DLVCs as good predictors for rectal bleeding. The established NTCP model based on these two DLVCs has been used routinely at our proton center during the plan evaluation to minimize the possible incidence of rectal bleeding in prostate cancer patients treated by IMPT. Table 1 Performances of the NTCP models.

Technical Note: Multiple energy extraction techniques for synchrotron-based proton delivery systems may exacerbate motion interplay effects in lung cancer treatments.

The multiple energy extraction (MEE) delivery technique for synchrotron-based proton delivery systems reduces beam delivery time by decelerating the beam multiple times during one accelerator spill, but this might cause additional plan quality degradation due to intrafractional motion. We seek to determine whether MEE causes significantly different plan quality degradation compared to single energy extraction (SEE) for lung cancer treatments due to the interplay effect. Ten lung cancer patients treated with IMPT at our institution were nonrandomly sampled based on a representative range of tumor motion amplitudes, tumor volumes, and respiratory periods. Dose-volume histogram (DVH) indices from single-fraction SEE and MEE four-dimensional (4D) dynamic dose distributions were compared using the Wilcoxon signed-rank test. Distributions of monitor units (MU) to breathing phases were investigated for features associated with plan quality degradation. SEE and MEE DVH indices were compared in fractionated deliveries of the worst-case patient treatment scenario to evaluate the impact of fractionation. There were no clinically significant differences in target mean dose, target dose conformity, or dose to organs-at-risk between SEE and MEE in single-fraction delivery. Three patients had significantly worse dose homogeneity with MEE compared to SEE (single-fraction mean D5% -D95% increased by up to 9.6% of prescription dose), and plots of MU distribution to breathing phases showed synchronization patterns with MEE but not SEE. However, after 30 fractions the patient in the worst-case scenario had clinically acceptable target dose homogeneity and coverage with MEE (mean D5% -D95% increased by 1% compared to SEE). For some patients with breathing periods close to the mean spill duration, MEE resulted in significantly worse single-fraction target dose homogeneity compared to SEE due to the interplay effect. However, this was mitigated by fractionation, and target dose homogeneity and coverage were clinically acceptable after 30 fractions with MEE.

Exploratory Investigation of Dose-Linear Energy Transfer (LET) Volume Histogram (DLVH) for Adverse Events Study in Intensity Modulated Proton Therapy (IMPT).

We proposed a novel tool-a dose linear energy transfer (LET)-volume histogram (DLVH)-and performed an exploratory study to investigate rectal bleeding in prostate cancer treated with intensity modulated proton therapy. The DLVH was constructed with dose and LET as 2 axes, and the normalized volume of the structure was contoured in the dose-LET plane as isovolume lines. We defined the DLVH index, DLv%(d,l) (ie, v% of the structure) to have a dose of ≥d Gy and an LET of ≥l keV/μm, similar to the dose-volume histogram index Dv%. Nine patients with prostate cancer with rectal bleeding (Common Terminology Criteria for Adverse Events grade ≥2) were included as the adverse event group, and 48 patients with no complications were considered the control group. A P value map was constructed by comparison of the DLVH indices of all patients between the 2 groups using the Mann-Whitney U test. Dose-LET volume constraints (DLVCs) were derived based on the P value map with a manual selection procedure facilitated by Spearman's correlation tests. The obtained DLVCs were further cross-validated using a multivariate support vector machine (SVM)-based normal tissue complication probability (NTCP) model with an independent testing data set composed of 8 adverse event and 13 control patients. We extracted 2 DLVC constraints. One DLVC was obtained, Vdose/LETboundary:2.5keVμmat 75 Gy to 3.2keVμmat8.65Gy <1.27% (DLVC1), revealing a high LET volume effect. The second DLVC, V(72.2Gy,0keVμm) < 2.23% (DVLC2), revealed a high dose volume effect. The SVM-based NTCP model with 2 DLVCs provided slightly superior performance than using dose only, with an area under the curve of 0.798 versus 0.779 for the testing data set. Our results demonstrated the importance of rectal "hot spots" in both high LET (DLVC1) and high dose (DLVC2) in inducing rectal bleeding. The SVM-based NTCP model confirmed the derived DLVCs as good predictors for rectal bleeding when intensity modulated proton therapy is used to treat prostate cancer.

Using Novel Statistical Techniques to Accurately Determine the Predictive Dose Range in a Study of Overall Survival after Definitive Radiotherapy for Stage III Non-Small Cell Lung Cancer in Association with Heart Dose.

Purpose:Recent studies of radiotherapy (RT) for stage III non-small-cell lung cancer (NSCLC) have associated high dose to the heart with cardiac toxicity and decreased overall survival (OS). We used advanced statistical techniques to account for correlations between dosimetric variables and more accurately determine the range of heart doses which are associated with reduced OS in patients receiving RT for stage III NSCLC.Methods:From 2006 to 2013, 119 patients with stage III NSCLC received definitive RT at our institution. OS data was obtained from institutional tumor registry. We used multivariate Cox model to determine patient specific covariates predictive for reduced overall survival. We examined age, prescription dose, mean lung dose, lung V20, RT technique, stage, chemotherapy, tumor laterality, tumor volume, and tumor site as candidate covariates. We subsequently used novel statistical techniques within multivariate Cox model to systematically search the whole heart dose-volume histogram (DVH) for dose parameters associated with OS.Results:Patients were followed until death or 2.5 to 81.2 months (median 30.4 months) in those alive at last follow up. On multivariate analysis of whole heart DVH, the dose of 51 Gy was identified as a threshold dose above which the dose volume relationship becomes predictive for OS. We identified V55Gy (percentage of the whole heart volume receiving at least 55 Gy) as the best single DVH index which can be used to set treatment optimization constraints (Hazard Ratio = 1.044 per 1% increase in heart volume exposed to at least 55 Gy, P = 0.03). Additional characteristics correlated with OS on multivariate analysis were age, stage (IIIA/IIIB), and administration of chemotherapy.Conclusion:Doses above 51 Gy, applied to small volumes of the heart, are associated with worse OS in stage III NSCLC patients treated with definitive RT. Higher stage, older age and lack of chemotherapy were also associated with reduced OS.

The benefits evaluation of abdominal deep inspiration breath hold based on knowledge-based radiotherapy treatment planning for left-sided breast cancer.

PurposeTo study the impact of abdominal deep inspiration breath hold (DIBH) technique on knowledge‐based radiotherapy treatment planning for left‐sided breast cancer to guide the application of DIBH technology.Materials and methodsTwo kernel density estimation (KDE) models were developed based on 40 left‐sided breast cancer patients with two CT acquisitions of free breathing (FB‐CT) and DIBH (DIBH‐CT). Each KDE model was used to predict dose volume histograms (DVHs) based on DIBH‐CT and FB‐CT for another 10 new patients similar to our training datasets. The predicted DVHs were taken as a substitute for dose constraints and objective functions in the Eclipse treatment planning system, with the same requirements for the planning target volume (PTV). The mean doses to the heart, the left anterior descending coronary artery (LADCA) and the ipsilateral lung were evaluated and compared using the T‐test among clinical plans, KDE predictions, and KDE plans.ResultsOur study demonstrated that the KDE model can generate deliverable simulations equivalent to clinically applicable plans. The T‐test was applied to test the consistency hypothesis on another ten left‐sided breast cancer patients. In cases of the same breathing status, there was no statistically significant difference between the predicted and the clinical plans for all clinically relevant DVH indices (P > 0.05), and all predicted DVHs can be transferred into deliverable plans. For DIBH‐CT images, significant differences were observed between FB model predictions and clinical plans (P < 0.05). DIBH model prediction cannot be optimized to a deliverable plan based on FB‐CT, with a counsel of perfection.ConclusionKDE models can predict DVHs well for the same breathing conditions but degrade with different breathing conditions. The benefits of DIBH for a given patient can be evaluated with a quick comparison of prediction results of the two models before treatment planning.

Evaluating the impact of real-time multicriteria optimizers integrated with interactive plan navigation tools for HDR brachytherapy

PurposeCurrently in high-dose-rate (HDR) brachytherapy planning, manual fine-tuning of an objective function is a common practice. Furthermore, automated planning approaches such as multicriteria optimization (MCO) are still limited to the automatic generation of a single treatment plan. This study aims to quantify planning efficiency gains when using a graphics processing unit–based MCO (gMCO) algorithm combined with a novel graphical user interface (gMCO-GUI) that integrates efficient automated and interactive plan navigation tools. Methods and MaterialsThe gMCO algorithm was used to generate 1000 Pareto optimal plans per case for 379 prostate cases. gMCO-GUI was developed to allow plan navigation through all plans. gMCO-GUI integrates interactive parameter selection tools directly with the optimization algorithm to allow plan navigation. The quality of each plan was evaluated based on the Radiation Treatment Oncology Group 0924 protocol and a more stringent institutional protocol (INSTp). gMCO-GUI allows real-time time display of the dose–volume histogram indices, the dose–volume histogram curves, and the isodose lines during the plan navigation. ResultsOver the 379 cases, the fraction of Radiation Treatment Oncology Group 0924 protocol valid plans with target coverage greater than 95% was 90.8%, compared with 66.0% for clinical plans. The fraction of INSTp valid plans with target coverage greater than 95% was 81.8%, compared with 62.3% for clinical plans. The average time to compute 1000 deliverable plans with gMCO was 12.5 s, including the full computation of the 3D dose distributions. ConclusionsCombining the gMCO algorithm with automated and interactive plan navigation tools resulted in simultaneous gains in both plan quality and planning efficiency.

Dose prediction using a deep neural network for accelerated planning of rectal cancer radiotherapy

PurposeTo apply a deep neural network to predict dose distributions of rectal cancer patients for accelerated volume modulated arc technique (VMAT) planning. Materials and methodsComputed tomography scans and approved VMAT plans together with Doseapproved of 187 patients treated from February 2018 to April 2019 were randomly selected for this retrospective study. The deep neural network DeepLabv3+ was applied for dose distribution prediction. A prior dose information-aided planning scheme was introduced. Prediction precision was evaluated by mean square error (MSE), normalized dose difference (δD), and dose–volume histogram (DVH) indices using a paired t test. Information-aided and experienced replanning were performed by 1-year and 6-year experienced dosimetrists, respectively. Replanning time and DVH indices were evaluated by two-way variance analysis. ResultsThe DeepLabv3+ prediction results (DoseDeepLabv3+) were all clinically acceptable. Taking Doseapproved as the baseline, the MSE was 0.001 and mean δD was 0.40% with an inter-quartile range of 0.079%–0.30% for DoseDeepLabv3+. No significant differences were found for the planning target volume quantitative parameters between Doseapproved and DoseDeepLabv3+, except for the conformality index. For the two-way variance analysis, a significantly different replanning time was found between the information-aided and experienced replanning with maximum time-saving of 15.76 min. Information-aided replans had the advantage of lower maximum dose, higher minimum dose, and lower homogeneity index, and the disadvantage of lower conformality index and higher machine unites with significant differences. ConclusionDeepLabv3+ successfully predicted rectal cancer dose distribution, and the predicted prior information helped save planning times for multi-level experienced dosimetrists.

Robust Optimization for Intensity Modulated Proton Therapy to Redistribute High Linear Energy Transfer from Nearby Critical Organs to Tumors in Head and Neck Cancer

We propose linear energy transfer (LET)-guided robust optimization in intensity modulated proton therapy for head and neck cancer. This method simultaneously considers LET and physical dose distributions of tumors and organs at risk (OARs) with uncertainties. Fourteen patients with head and neck cancer were included in this retrospective study. Cord, brain stem, brain, and oral cavity were considered. Two algorithms, voxel-wise worst-case robust optimization and LET-guided robust optimization (LETRO), were used to generate intensity modulated proton therapy plans for each patient. The latter method directly optimized LET distributions rather than indirectly as in previous methods. LET-volume histograms (LETVHs) were generated, and high LET was redistributed from nearby OARs to tumors in a user-defined way via LET-volume constraints. Dose-volume histogram indices, such as clinical target volume (CTV) D98% and D2%-D98%, cord Dmax, brain stem Dmax, brain Dmax, and oral cavity Dmean, were calculated. Plan robustness was quantified using the worst-case analysis method. LETVH indices analogous to dose-volume histogram indices were used to characterize LET distributions. The Wilcoxon signed rank test was performed to measure statistical significance. In the nominal scenario, LETRO provided higher LET distributions in the CTV (unit: keV/μm; CTV LET98%: 1.18 vs 1.08, LETRO vs RO, P = .0031) while preserving comparable physical dose and plan robustness. LETRO achieved significantly reduced LET distributions in the cord, brain stem, and oral cavity compared with RO (cord LETmax: 7.20 vs 8.20, P=.0010; brain stem LETmax: 10.95 vs 12.05, P = .0007; oral cavity LETmean: 2.11 vs 3.12, P = .0052) and had comparable physical dose and plan robustness in all OARs. In the worst-case scenario, LETRO achieved significantly higher LETmean in the CTV, reduced LETmax in the brain, and was comparable to other LETVH indices (CTV LETmean: 3.26 vs 3.35, P= .0012; brain LETmax: 24.80 vs 22.00, P = .0016). LETRO robustly optimized LET and physical dose distributions simultaneously. It redistributed high LET from OARs to targets with slightly modified physical dose and plan robustness.