IMRT Optimization Research Articles

MO-FG-CAMPUS-TeP2-01: A Graph Form ADMM Algorithm for Constrained Quadratic Radiation Treatment Planning

Purpose: In radiation therapy optimization the constraints can be either hard constraints which must be satisfied or soft constraints which are included but do not need to be satisfied exactly. Currently the voxel dose constraints are viewed as soft constraints and included as a part of the objective function and approximated as an unconstrained problem. However in some treatment planning cases the constraints should be specified as hard constraints and solved by constrained optimization. The goal of this work is to present a computation efficiency graph form alternating direction method of multipliers (ADMM) algorithm for constrained quadratic treatment planning optimization and compare it with several commonly used algorithms/toolbox. Method: ADMM can be viewed as an attempt to blend the benefits of dual decomposition and augmented Lagrangian methods for constrained optimization. Various proximal operators were first constructed as applicable to quadratic IMRT constrained optimization and the problem was formulated in a graph form of ADMM. A pre-iteration operation for the projection of a point to a graph was also proposed to further accelerate the computation. Result: The graph form ADMM algorithm was tested by the Common Optimization for Radiation Therapy (CORT) dataset including TG119, prostate, liver, and head & neck cases. Both unconstrained and constrained optimization problems were formulated for comparison purposes. All optimizations were solved by LBFGS, IPOPT, Matlab built-in toolbox, CVX (implementing SeDuMi) and Mosek solvers. For unconstrained optimization, it was found that LBFGS performs the best, and it was 3–5 times faster than graph form ADMM. However, for constrained optimization, graph form ADMM was 8 – 100 times faster than the other solvers. Conclusion: A graph form ADMM can be applied to constrained quadratic IMRT optimization. It is more computationally efficient than several other commercial and noncommercial optimizers and it also used significantly less computer memory.

SU‐F‐T‐428: An Optimization‐Based Commissioning Tool for Finite Size Pencil Beam Dose Calculations

Purpose:Finite size pencil beam (FSPB) algorithms are commonly used to pre‐calculate the beamlet dose distribution for IMRT treatment planning. FSPB commissioning, which usually requires fine tuning of the FSPB kernel parameters, is crucial to the dose calculation accuracy and hence the plan quality. Yet due to the large number of beamlets, FSPB commissioning could be very tedious. This abstract reports an optimization‐based FSPB commissioning tool we have developed in MatLab to facilitate the commissioning.Methods:A FSPB dose kernel generally contains two types of parameters: the profile parameters determining the dose kernel shape, and a 2D scaling factors accounting for the longitudinal and off‐axis corrections. The former were fitted using the penumbra of a reference broad beam's dose profile with Levenberg‐Marquardt algorithm. Since the dose distribution of a broad beam is simply a linear superposition of the dose kernel of each beamlet calculated with the fitted profile parameters and scaled using the scaling factors, these factors could be determined by solving an optimization problem which minimizes the discrepancies between the calculated dose of broad beams and the reference dose.Results:We have commissioned a FSPB algorithm for three linac photon beams (6MV, 15MV and 6MVFFF). Dose of four field sizes (6*6cm2, 10*10cm2, 15*15cm2 and 20*20cm2) were calculated and compared with the reference dose exported from Eclipse TPS system. For depth dose curves, the differences are less than 1% of maximum dose after maximum dose depth for most cases. For lateral dose profiles, the differences are less than 2% of central dose at inner‐beam regions. The differences of the output factors are within 1% for all the three beams.Conclusion:We have developed an optimization‐based commissioning tool for FSPB algorithms to facilitate the commissioning, providing sufficient accuracy of beamlet dose calculation for IMRT optimization.

SU‐F‐T‐257: Comparison Study of the Biological and Physical Optimization for Cervical Carcinoma

Purpose:In this study, IMRT optimization based on biological related models were applied in the clinic for cervical carcinoma cases and the results were compared with the traditional physical optimization. Methods: Twenty patients of advanced stage cervical carcinoma were selected in this study. The treatment plan was generated on each patient using both biological and physical optimized model respectively. Additionally, the biological optimization was based on two basic models for bladder and rectum: Poisson (adjustable parameters: alpha/beta, gamma, D50) and Lyman‐Kutcher‐Burman (adjustable parameters: m, n, D50). And then each optimized biological plan was compared with the corresponding physical plan. During this comparison, all biological and physical parameters, such as NTCP,TCP, target coverage and homogeneous index(HI), were evaluated by paired T‐test to verify IMRT plans quality.Results:The NTCP values by biological optimization were lower than the values from corresponding physical optimization. For the bladder and rectum, the detail of NTCP values are 0.144 and 0.178 with p=0.028 for bladder, and 0.108 and 0.186 with p= 7.97E‐5 for rectum respectively. The mean dose and V40 of bladder and rectum were comparable, and there was no significant difference between these two optimizations. For PTV, both optimized models achieved similar target coverage and homogeneous index (For example, D98=53.4 Gy and 54.6 Gy from biological optimization and physical optimization respectively with p=2.94E‐9), and there were no significant differences in the tumor control probability(TCP) values.Conclusion:IMRT optimization based on biological model did not decrease the plan quality, and could spare bladder and rectum with lower NTCP value much more compared with traditional physical optimization for cervical carcinoma. Our initial experiment demonstrated that the biological optimization could be a potential alternative method to the traditional physical one.We want to thank Dr. Lei Xing and Dr. Yong Yang in the Stanford University School of Medicine for this work. This work was jointly supported by NSFC (61471226), Natural Science Foundation for Distinguished Young Scholars of Shandong Province (JQ201516), and China Postdoctoral Science Foundation (2015T80739, 2014M551949).

SU‐F‐T‐416: Dosimetric Comparison of Coplanar and Non‐Coplanar IMRT Plans for Peripheral Lung Lesion

Purpose:The purpose of this study was to compare dosimetric parameters of treatment plans between coplanar and non‐coplanar techniques for treating peripheral lung lesions.Methods:The planning CT scans of 6 patients in supine positions were used in this study. The size of the PTV ranges from 163 c.c. to 782 c.c.. The locations of PTV are mostly at the peripheral of Lung, some spreading to the mediastinum. For each patient, we generated two IMRT plans, one with and the other without non‐coplanar beams. The non‐coplanar beams were carefully selected so that the beams would never exit patient bodies through the contralateral lung. The IMRT plans were generated with Pinnacle 9.8 treatment planning software. The IMRT optimization objectives were kept the same for the corresponding pairs of plans. All plans were normalized such that 95% of PTV receives the prescription dose (full dose).Results:The conformity index (mean±standard deviation of the mean) is 1.49±0.14 and 1.58±0.23 for the coplanar and noncoplanar plans, respectively. The heterogeneity index (mean±standard deviation of the mean) is 7.74 ±2.33 and 6.34±1.40 for the coplanar and non‐coplanar plans, respectively. The maximum heart dose is 60.94±6.22 and 60.42±7.21 Gy, and mean heart dose is 10.22 ±7.57, 9.07 ±6.32 Gy, for the coplanar and non‐coplanar plans, respectively. The ipsilateral lung V20 is 48.0%±2.4% and 47.5%±3.3%, and V5 is 68.2%±10.0% and 69.1%±7.3%, for the coplanar and noncoplanar plans, respectively. Furthermore, with the non‐coplanar beam arrangement, the contralateral lung V20 was reduced from 3.3%±3.7% to 1.3%±0.8%, and the contralateral Lung V5 is reduced significantly from 65.6%±9.3% to 33.5%±20.9% (p value =0.008).Conclusion:The IMRT plans with non‐coplanar beam arrangement could reduce the exit dose to the contralateral lung, and therefore reduce the contralateral lung V5 significantly. This method is especially helpful while the lung lesion doesn't have a symmetric shape.

EP-1627: Knowledge-based IMRT optimisation using a model trained with VMAT plans of other setup orientations

EP-1627: Knowledge-based IMRT optimisation using a model trained with VMAT plans of other setup orientations

Personalized treatment planning with a model of radiation therapy outcomes for use in multiobjective optimization of IMRT plans for prostate cancer.

PurposeTo build a new treatment planning approach that extends beyond radiation transport and IMRT optimization by modeling the radiation therapy process and prognostic indicators for more outcome-focused decision making.MethodsAn in-house treatment planning system was modified to include multiobjective inverse planning, a probabilistic outcome model, and a multi-attribute decision aid. A genetic algorithm generated a set of plans embodying trade-offs between the separate objectives. An influence diagram network modeled the radiation therapy process of prostate cancer using expert opinion, results of clinical trials, and published research. A Markov model calculated a quality adjusted life expectancy (QALE), which was the endpoint for ranking plans.ResultsThe Multiobjective Evolutionary Algorithm (MOEA) was designed to produce an approximation of the Pareto Front representing optimal tradeoffs for IMRT plans. Prognostic information from the dosimetrics of the plans, and from patient-specific clinical variables were combined by the influence diagram. QALEs were calculated for each plan for each set of patient characteristics. Sensitivity analyses were conducted to explore changes in outcomes for variations in patient characteristics and dosimetric variables. The model calculated life expectancies that were in agreement with an independent clinical study.ConclusionsThe radiation therapy model proposed has integrated a number of different physical, biological and clinical models into a more comprehensive model. It illustrates a number of the critical aspects of treatment planning that can be improved and represents a more detailed description of the therapy process. A Markov model was implemented to provide a stronger connection between dosimetric variables and clinical outcomes and could provide a practical, quantitative method for making difficult clinical decisions.

FLT PET Quantification of Incomplete Pelvic Bone Marrow Recovery After Chemoradiation

To quantify the long-term effects of chemoradiation therapy on pelvic bone marrow with FLT PET imaging. Ten pelvic cancer patients were enrolled and completed an IRB-approved protocol designed to obtain FLT PET/CT images prior to chemoradiation therapy and 30 days and 1 year after chemotherapy was completed. All FLT PET images were registered to the radiation therapy planning CT by first registering the attenuation correction CT using the pelvic bones, and then copying the transformation matrix to the corresponding FLT PET image. Pretherapy images were used as a control for pelvic bone marrow FLT uptake change and to create active bone marrow avoidance regions in IMRT optimization. Radiation therapy plans were created to spare as much active bone marrow identified by the pretherapy FLT PET image as possible without compromising target dose or other normal tissue objectives. In order to analyze bone marrow activity changes as measured by FLT PET images, the dose distributions for each subject were divided into 1 Gy/week dose volume bins within the bony pelvis. The mean FLT PET standardized uptake value (SUV) was then recorded from all images in each dose bin at each time point. Complete blood counts were also collected for each subject at the time of each FLT PET scan. For the ten subjects imaged that completed all five FLT PET images, apparent recovery seen at 30 days was lost at 1 year as measured by FLT uptake in the regions that received radiation doses >25 Gy. On average, bone marrow activity as measured by mean FLT SUV returned to 56.4% ± 17.7% and 54.0% ± 15.5% of its pretherapy level in regions that received 20 – 25 Gy at 30 days and 1 year post therapy, respectively. The percentage of baseline FLT uptake was within 3% or slightly improved at 1 year in bone marrow receiving 25 Gy or less when compared to 30 day images. In regions that received >35 Gy, the percentage of baseline FLT uptake at 30 days was 18.8% ± 1.8% greater than that at 1 year. This pattern of recovery was also seen in CBC metrics. At 30 days, WBCs, platelets, and lymphocytes were statistically significantly lower than baseline values (P < .05, paired t-test). At 1 year; WBCs, platelets, neutrophils, and lymphocytes values were all statistically significantly lower than baseline values (P < .05, paired t-test). FLT PET imaging data measuring the long-term effects of chemoradiation therapy on pelvic bone marrow may indicate that apparent bone marrow recovery 30 days after therapy is not sustained 1 year after therapy in those regions receiving doses more than 25 Gy. Recovery sustained at 1 year is incomplete and inversely proportional to radiation dose.

Dosimetric impact of different multileaf collimators on prostate intensity modulated treatment planning

AimThe main purpose of this study is to perform a dosimetric comparison on target volumes and organs at risks (OARs) between prostate intensity modulated treatment plans (IMRT) optimized with different multileaf collimators (MLCs). BackgroundThe use of MLCs with a small leaf width in the IMRT optimization may improve conformity around the tumor target whilst reducing the dose to normal tissues. Materials and methodsTwo linacs mounting MLCs with 5 and 10mm leaf-width, respectively, implemented in Pinnacle3 treatment planning system were used for this work. Nineteen patients with prostate carcinoma undergoing a radiotherapy treatment were enrolled. Treatment planning with different setup arrangements (7 and 5 beams) were performed for each patient and each machine. Dose volume histograms (DVHs) cut-off points were used in the treatment planning comparison. ResultsComparable planning target volume (PTV) coverage was obtained with 7- and 5-beam configuration (both with 5 and 10mm MLC leaf-width). The comparison of bladder and rectum DVH cut-off points for the 5-beam arrangement shows that 52.6% of the plans optimized with a larger leaf-width did not satisfy at least one of the OARs’ constraints. This percentage is reduced to 10.5% for the smaller leaf-width. If a 7-beam arrangement is used the value of 52.6% decreases to 21.1% while the value of 10.5% remains unchanged. ConclusionMLCs collimators with different widths and number of leaves lead to a comparable prostate treatment planning if a proper adjustment is made of the number of gantry angles.

SU‐E‐T‐355: Efficient Scatter Correction for Direct Ray‐Tracing Based Dose Calculation

Purpose:To propose a scatter correction method with linear computational complexity for direct‐ray‐tracing (DRT) based dose calculation. Due to its speed and simplicity, DRT is widely used as a dose engine in the treatment planning system (TPS) and monitor unit (MU) verification software, where heterogeneity correction is applied by radiological distance scaling. However, such correction only accounts for attenuation but not scatter difference, causing the DRT algorithm less accurate than the model‐based algorithms for small field size in heterogeneous media.Methods:Inspired by the convolution formula derived from an exponential kernel as is typically done in the collapsed‐cone‐convolution‐superposition (CCCS) method, we redesigned the ray tracing component as the sum of TERMA scaled by a local deposition factor, which is linear with respect to density, and dose of the previous voxel scaled by a remote deposition factor, D(i)=aρ(i)T(i)+(b+c(ρ(i)‐1))D(i‐1),where T(i)=e^(‐αr(i)+β(r(i))^2) and r(i)=Σ_(j=1,..,i)ρ(j).The two factors together with TERMA can be expressed in terms of 5 parameters, which are subsequently optimized by curve fitting using digital phantoms for each field size and each beam energy.Results:The proposed algorithm was implemented for the Fluence‐Convolution‐Broad‐Beam (FCBB) dose engine and evaluated using digital slab phantoms and clinical CT data. Compared with the gold standard calculation, dose deviations were improved from 20% to 2% in the low density regions of the slab phantoms for the 1‐cm field size, and within 2% for over 95% of the volume with the largest discrepancy at the interface for the clinical lung case.Conclusion:We developed a simple recursive formula for scatter correction for the DRT‐based dose calculation with much improved accuracy, especially for small field size, while still keeping calculation to linear complexity. The proposed calculator is fast, yet accurate, which is crucial for dose updating in IMRT optimization.

SU‐E‐T‐500: Initial Implementation of GPU‐Based Particle Swarm Optimization for 4D IMRT Planning in Lung SBRT

Purpose4D‐IMRT planning, combined with dynamic MLC tracking delivery, utilizes the temporal dimension as an additional degree of freedom to achieve improved OAR‐sparing. The computational complexity for such optimization increases exponentially with increase in dimensionality. In order to accomplish this task in a clinically‐feasible time frame, we present an initial implementation of GPU‐based 4D‐IMRT planning based on particle swarm optimization (PSO).MethodsThe target and normal structures were manually contoured on ten phases of a 4DCT scan of a NSCLC patient with a 54cm3 right‐lower‐lobe tumor (1.5cm motion). Corresponding ten 3D‐IMRT plans were created in the Eclipse treatment planning system (Ver‐13.6). A vendor‐provided scripting interface was used to export 3D‐dose matrices corresponding to each control point (10 phases × 9 beams × 166 control points = 14,940), which served as input to PSO. The optimization task was to iteratively adjust the weights of each control point and scale the corresponding dose matrices. In order to handle the large amount of data in GPU memory, dose matrices were sparsified and placed in contiguous memory blocks with the 14,940 weight‐variables. PSO was implemented on CPU (dual‐Xeon, 3.1GHz) and GPU (dual‐K20 Tesla, 2496 cores, 3.52Tflops, each) platforms. NiftyReg, an open‐source deformable image registration package, was used to calculate the summed dose.ResultsThe 4D‐PSO plan yielded PTV coverage comparable to the clinical ITV‐based plan and significantly higher OAR‐sparing, as follows: lung Dmean=33%; lung V20=27%; spinal cord Dmax=26%; esophagus Dmax=42%; heart Dmax=0%; heart Dmean=47%. The GPU‐PSO processing time for 14940 variables and 7 PSO‐particles was 41% that of CPU‐PSO (199 vs. 488 minutes).ConclusionTruly 4D‐IMRT planning can yield significant OAR dose‐sparing while preserving PTV coverage. The corresponding optimization problem is large‐scale, non‐convex and computationally rigorous. Our initial results indicate that GPU‐based PSO with further software optimization can make such planning clinically feasible.This work was supported through funding from the National Institutes of Health and Varian Medical Systems.

SU‐E‐T‐435: Flattening Filter Free Beams for Head and Neck IMRT and VMAT Optimization

Purpose:To investigate if there is benefits of using flattening filter free (FFF) versus flattening filter (FF) beams during plan optimization for head and neck targets.Methods:Five head and neck cancer patients previously treated were selected for this study. Each plan was optimized for volumetric modulated arc therapy (VMAT) using 2 full arcs. Target volumes and the organs at risk (OAR) were outlined by the same physician. The prescription doses for each patient and the dose limits for each OAR were specified by the physician. For each patient four plans were created by varying the photon beam energy (6MV vs 10MV) and the use of FF or FFF. For each patient the prescription dose and OAR dose limits remained unchanged. Furthermore, the number of optimization iterations remained the same for each plan without user interference. The plans were compared using dose volume histograms, conformity and homogeneity indices.Results:Our results showed that the plans optimized with FF beams produced more homogeneous dose distributions in the PTV than the ones with FFF beams (0.92 vs. 0.86 and 0.92 vs.85 for 6MVFF, 6MVFFF, 10MVFF and 10MVFFF respectively). The conformity was better for FFF beams than for FF beams (1.26 vs. 1.20 and 1.27 vs. 1.21 for 6MVFF, 6MVFFF, 10MVFF and 10MVFFF respectively). The overall minimum, maximum and means doses were reduced for the FFF plans by approximately 4% to 6%. All FFF plans allowed more sparing of organs at risk by 7% to 9% on average. Only one 10MVFFF plan had a 1% higher dose than the respective 10MVFF for the spinal cord.Conclusion:The 6MVFFF and 10MVFFF beams illustrated the improvement of normal tissue sparing while achieving similar target dose distribution compared to 6MVFF and 10MVFF beams. Further investigation is required to determine the magnitude of improvement. case specific, fluence‐based optimization improved both PTV and OAR dosimetry in 80% of cases.

SU‐E‐T‐218: Comprehensive Plan Integrity and Quality Check by Accessing Eclipse Planning Data Remotely Via a Novel Eclipse‐API Client‐Server Interface

Purpose:It is desirable to seamlessly and remotely access treatment plan data in Eclipse in order to allow sequential treatment planning and quality assurance automation, e.g. automatic plan quality check and comprehensive plan integrity check. The manual DICOM export function and the native Eclipse SQL interface cannot fully support the needs. Therefore we have developed a procedure and computer programs to enable access of Eclipse planning data remotely, and directly from Eclipse server, via Eclipse API, so to support such clinical software applications requiring advanced mathematics and extensive computation.Methods:A novel client‐server interface was developed to allow external software programs to remotely and seamlessly access the Eclipse planning data using Eclipse‐API from outside the Eclipse workstations. Remote (client) programs send requests as network socket messages to the server program running on an Eclipse workstation, and the server program uses Eclipse‐API functions to access planning data, and sends data back to the client programs.Results:The Eclipse‐API client‐server interface were implemented and successfully tested in clinical settings. Comprehensive plan integrity and quality check programs, which were previously supporting Pinnacle TPS only, have been extended to support Eclipse and are executable on any network computer in the department independent of Eclipse installation. Advanced plan check features, e.g. IMRT optimization structure overlapping check, on‐the‐fly derivation of new structures in planning quality check, that were previously not possible in C# programs via Eclipse‐API, have now been implemented, allowed by the inclusion of advanced mathematics packages e.g. MATLAB in the external programs.Conclusion:The new Eclipse‐API client‐server interface has enabled Eclipse data to be accessed remotely and seamlessly. It has the potential to significantly improve the comprehensiveness of third party software programs based on Eclipse‐API for both clinical and research purposes, and would be useful to extend the Eclipse / ARIA ecosystem.This work was partially funded by Varian Medical System

SU-E-J-68: Adaptive Radiotherapy of Head and Neck Cancer: Re-Planning Based On Prior Dose

Purpose: Adaptive Radiotherapy (ART) with frequent CT imaging has been used to improve dosimetric accuracy by accounting for anatomical variations, such as primary tumor shrinkage and/or body weight loss, in Head and Neck (H&N) patients. In most ART strategies, the difference between the planned and the delivered dose is estimated by generating new plans on repeated CT scans using dose-volume constraints used with the initial planning CT without considering already delivered dose. The aim of this study was to assess the dosimetric gains achieved by re-planning based on prior dose by comparing them to re-planning not based-on prior dose for H&N patients. Methods: Ten locally-advanced H&N cancer patients were selected for this study. For each patient, six weekly CT imaging were acquired during the course of radiotherapy. PTVs, parotids, cord, brainstem, and esophagus were contoured on both planning and six weekly CT images. ART with weekly re-plans were done by two strategies: 1) Generating a new optimized IMRT plan without including prior dose from previous fractions (NoPriorDose) and 2) Generating a new optimized IMRT plan based on the prior dose given from previous fractions (PriorDose). Deformable image registration was used to accumulate the dose distributions between planning and six weekly CT scans. The differences in accumulated doses for both strategies were evaluated using the DVH constraints for all structures. Results: On average, the differences in accumulated doses for PTV1, PTV2 and PTV3 for NoPriorDose and PriorDose strategies were <2%. The differences in Dmean to the cord and brainstem were within 3%. The esophagus Dmean was reduced by 2% using PriorDose. PriorDose strategy, however, reduced the left parotid D50 and Dmean by 15% and 14% respectively. Conclusion: This study demonstrated significant parotid sparing, potentially reducing xerostomia, by using ART with IMRT optimization based on prior dose for weekly re-planning of H&N cancer patients.

SU‐E‐P‐47: Evaluation of Improvement of Esophagus Sparing in SBRT Lung Patients with Biologically Based IMRT Optimization

Purpose:To study the potential of improving esophageal sparing for stereotactic body radiation therapy (SBRT) lung cancer patients by using biological optimization (BO) compared to conventional dose‐volume based optimization (DVO) in treatment planning.Methods:Three NSCLC patients (PTV (62.3cc, 65.1cc, and 125.1cc) adjacent to the heart) previously treated with SBRT were re‐planned using Varian Eclipse TPS (V11) using DVO and BO. The prescription dose was 60 Gy in 5 fractions normalized to 95% of the PTV volume. Plans were evaluated by comparing esophageal maximum doses, PTV heterogeneity (HI= D5%/D95%), and Paddick's conformity (CI) indices. Quality of the plans was assessed by clinically‐used IMRT QA procedures.Results:By using BO, the maximum dose to the esophagus was decreased 1384 cGy (34.6%), 502 cGy (16.5%) and 532 cGy (16.2%) in patient 1, 2 and 3 respectively. The maximum doses to spinal cord and the doses to 1000 cc and 1500 cc of normal lung were comparable in both plans. The mean doses (Dmean_hrt) and doses to 15cc of the heart (V15_hrt) were comparable for patient 1 and 2. However for patient 3, with the largest PTV, Dmean_hrt and V15_hrt increased by 62.2 cGy (18.3%) and 549.9 cGy (24.9%) respectively for the BO plans. The mean target HI of BO plans (1.13) was inferior to the DVO plans (1.07). The same trend was also observed for mean CI in BO plans (0.77) versus DVO plans (0.83). The QA pass rates (3%, 3mm) were comparable for both plans.Conclusion:This study demonstrated that the use of biological models in treatment planning optimization can substantially improve esophageal sparing without compromising spinal cord and normal lung doses. However, for the large PTV case (125.1cc) we studied here, Dmean_hrt and V15_hrt increased substantially. The target HI and CI were inferior in the BO plans.

TU-AB-303-01: A Feasibility Study for Dynamic Adaptive Therapy of Non-Small Cell Lung Cancer

Purpose: To compare plans for NSCLC optimized using Dynamic Adaptive Therapy (DAT) with conventional IMRT optimization. DAT adapts plans based on changes in the target volume by using dynamic programing techniques to consider expected changes into the optimization process. Information gathered during treatment, e.g. from CBCT, is incorporated into the optimization. Methods and materials: DAT is formulated using stochastic control formalism, which minimizes the total expected number of tumor cells at the end of a treatment course subject to uncertainty inherent in the tumor response and organs-at-risk (OAR) dose constraints. This formulation allows for non-stationary dose distribution as well as non-stationary fractional dose as needed to achieve a series of optimal plans that are conformal to tumor over time. Sixteen phantom cases with various sizes and locations of tumors, and OAR geometries were generated. Each case was planned with DAT and conventional IMRT (60Gy/30fx). Tumor volume change over time was obtained by using, daily MVCT-based, two-level cell population model. Monte Carlo simulations have been performed for each treatment course to account for uncertainty in tumor response. Same OAR dose constraints were applied for both methods. The frequency of plan modification was varied to 1, 2, 5 (weekly), and 29 (daily). The final average tumor dose and OAR doses have been compared to quantify the potential benefit of DAT. Results: The average tumor max, min, mean, and D95 resulted from DAT were 124.0–125.2%, 102.1–114.7%, 113.7–123.4%, and 102.0–115.9% (range dependent on the frequency of plan modification) of those from conventional IMRT. Cord max, esophagus max, lung mean, heart mean, and unspecified tissue D05 resulted from AT were 84–102.4%, 99.8–106.9%, 66.9–85.6%, 58.2–78.8%, and 85.2–94.0% of those from conventional IMRT. Conclusions: Significant tumor dose increase and OAR dose reduction, especially with parallel OAR with mean or dose-volume constraints, can be achieved using DAT.

MO‐D‐213‐04: The Proximity to the Skin of PTV Affects PTV Coverage and Skin Dose for TomoTherapy

Purpose:The proximity to the skin surface of the PTV for the patients with skin disease could be a concern in terms of the PTV coverage and actual surface dose (SD). IMRT optimization algorithms increase the beam intensity close to the skin in order to compensate for lack of scattering material, leading to enhanced SD but potential hot spots. This study aims to investigate the effect of PTV proximity to the skin on planning and measured SDMethods:All measurements were done for 6 MV X‐ray beam of Helical TomoTherapy. An anthropomorphic phantom was scanned in a CT simulator in a routine manner with thermoplastic mask immobilization. PTVs were created with varying distances to the skin of 0 mm ‐(PTV1), 1 mm‐ (PTV2), 2 mm‐(PTV3) and 3 mm‐(PTV4). Also, a 5 mm bolus was used with PTV1 (PTV5). All planning constraints were kept the same in all studies (hard constraint: 95% of the prescription dose covered 95% of the PTV). Gafchromic film (EBT3) was placed under the mask on the phantom surface, and the resulting dose was estimated using RIT software.Results:Optimizing the dose using different PTVs lead to average planned target doses of 10.8, 10.3, 10.2, 10.3 and 10.0 Gy, with maximum doses 12.2, 11.2, 11.1, 11.1 and 10.0 Gy for PTV1, PTV2, PTV3, PTV4 and PTV5, respectively. EBT3 measurements indicated a significant decrease of SD with skin distance by 12.7% (PTV1), 21.9% (PTV2), 24.8% (PTV3) and 28.4% (PTV4) comparing to prescription dose. Placement of a 5 mm bolus on the phantom surface resulted in a SD close to prescribed (+0.5%).Conclusion:This work provides a clear demonstration of the relationship between the skin dose and the PTV to the skin distance. The results indicate the necessity of a bolus even for TomoTherapy when high skin dose is required.

TH‐AB‐BRB‐01: Solving Larger IMRT Problems by Enhanced Reduced‐Order Constrained Optimization

Purpose:To report the latest improvements in the pipeline of the reduced‐order constrained optimization (ROCO) method for fast IMRT planning.Methods:The ROCO method involves the characterization of the solution space of the underlying fluence map IMRT optimization problem in a compact form, which enables forming a clinically tractable constrained optimization problem. The offline part of the method consists of three stages: sampling, learning the effective mode space, and mode‐dose calculation. In the sampling stage, a set of unconstrained optimization problems is solved, each corresponding to a different choice of weights and dose‐volume limits. As opposed to previous implementations, we preprocess the samples, identify the effective beamlets with non‐zero variances, and use them to attain a compact representation of the fluence space using Principal Component Analysis (PCA). In the mode‐dose calculation stage, the dose corresponding to each mode is computed based on the underlying full dose‐calculation technique. We offer a new formulation for handling possible negative mode elements, exploiting linearity of the mode‐dose map. In the online phase, a constrained optimization problem with a minimal number of constraints is solved over the basis coefficients spanning the reduced‐size space.Results:The results suggest that for a typical prostate case, the PCA stage can be performed about 5 times faster. Moreover, the mode dose calculation performance is accelerated by a factor of 2. Apart from computational efficiency, we also observed significantly lower memory demand for ROCO model compared to optimization models relying on pre‐calculation of dose deposition coefficients (DDCs). The proposed method has been integrated into the Varian Eclipse version 13.5 treatment planning system as a stand‐alone application.Conclusion:The new implementation has led a substantial savings in memory requirements and computational effort, allowing larger IMRT problems with finer beamlet resolution to be addressed with the available memory resources in less time.

First application of quantum annealing to IMRT beamlet intensity optimization

Optimization methods are critical to radiation therapy. A new technology, quantum annealing (QA), employs novel hardware and software techniques to address various discrete optimization problems in many fields. We report on the first application of quantum annealing to the process of beamlet intensity optimization for IMRT.We apply recently-developed hardware which natively exploits quantum mechanical effects for improved optimization. The new algorithm, called QA, is most similar to simulated annealing, but relies on natural processes to directly minimize a system’s free energy. A simple quantum system is slowly evolved into a classical system representing the objective function. If the evolution is sufficiently slow, there are probabilistic guarantees that a global minimum will be located.To apply QA to IMRT-type optimization, two prostate cases were considered. A reduced number of beamlets were employed, due to the current QA hardware limitations. The beamlet dose matrices were computed using CERR and an objective function was defined based on typical clinical constraints, including dose-volume objectives, which result in a complex non-convex search space. The objective function was discretized and the QA method was compared to two standard optimization methods, simulated annealing and Tabu search, run on a conventional computing cluster.Based on several runs, the average final objective function value achieved by the QA was 16.9 for the first patient, compared with 10.0 for Tabu and 6.7 for the simulated annealing (SA) method. For the second patient, the values were 70.7 for the QA, 120.0 for Tabu and 22.9 for the SA. The QA algorithm required 27–38% of the time required by the other two methods.In this first application of hardware-enabled QA to IMRT optimization, its performance is comparable to Tabu search, but less effective than the SA in terms of final objective function values. However, its speed was 3–4 times faster than the other two methods. This initial experiment suggests that more research into QA-based heuristics may offer significant speedup over conventional clinical optimization methods, as quantum annealing hardware scales to larger sizes.

IMPACT OF SEGMENTS NUMBER REDUCTION IN IMRT PLANNING

Purpose: To investigate the planning and dosimetric advantages of direct aperture optimization (DAO) over beamlet optimization in IMRT treatment of head and neck (H/N) and prostate cancers. &#x0D; Materials and methods: a brain metastatic case with multiple lesions, five head and neck, as well as five prostate patients were planned using the beamlet optimizer in Elekta-Xio© ver 4.6 IMRT treatment planning system. Based on our experience in beamlet IMRT optimization, PTVs in brain were prescribed to 66Gy using 5 fields, PTVs in H/N plans were prescribed to 70 Gy delivered by 7 fields, and prostate PTVs were prescribed to 76 Gy using nine fields. In all plans, fields were set to be equally spaced. All cases were re-planed using Direct Aperture optimizer (DAO) in Prowess Panther© ver 5.01 IMRT planning system at same configurations and dose constraints. Plans were evaluated according to ICRU criteria, number of segments, number of monitor units and planning time. &#x0D; Results: In brain case, beamlet optimization was better than DAO for both GTVs and PTVs in 95% isodose coverage, and the hot area was about 7% more in beamlet plan than DAO plan. For OAR, results showed an improvement in OAR sparing up to more than 35% in rt. eye, lt. eye, rt. optic nerve and lt. optic nerve when using DAO for planning, while optic chiasma sparing was about 20% also in beamlet optimizer plan. &#x0D; For H/N plans, the near maximum dose (D2) and the dose that covers 95% (D95) of PTV has improved by 4% in DAO. For organs at risk (OAR), DAO reduced the volume covered by 30% (V30) inspinal cord, right parotid, and left parotid by 60%, 54%, and 53% respectively. This considerable dosimetric quality improvement achieved using 25% less planning time and lower number of segments and monitor units by 46% and 51% respectively. &#x0D; In DAO prostate plans, Both D2 and D95 for the PTV wereimproved by only 2%. The V30 of right femur, left femur and bladder were improved by 35%, 15% and 3% respectively. On the contrary, the rectum V30 got even worse by 9%. However, number of monitor units, and number of segments decreased by 20% and 25% respectively. Moreover the planning time reduced significantly too. &#x0D; Conclusion: DAO introduces considerable advantages over beamlet optimization for different sites of cancer, in regards to organ at risk sparing. While no significant improvement occurred in the PTV ICRU reporting dose.

Poster - Thur Eve - 73: Follow up of a new bladder preparation protocol when treating the prostate

Background and Purpose: Patients treated for their prostate cancer were put on a bladder preparation protocol to obtain a full bladder. The associated high degree of discomfort forced us to change our practice. The patients were asked to drink 300ml prior to their treatment instead of 750ml. Following this change, a retrospective study was undertaken to assess the dosimetric impact on the bladder. Methods: The first group (patients who drank 750ml) had 10 patients. The second group (patients who drank 300ml) had 14 patients. Planning CT and 8 daily CBCT bladder volumes and DVHs were compared for both groups. Dose optimization constraints where based on PROFIT 78Gy/39 fraction arm. Results: For patients who drank 750ml, the average bladder volume was 377.1±60.9 cc compared with 236.5±31.5 cc for patients who drank 300 ml (p=0.043). Therapists do not report any case of incontinence in the 300 ml group. The achieved bladder wall constraint for V68% went from 23.7±1.7% to 33.8±2.0%, p=0.012, and the V91% went from 14.2±1.1% to 21.8±1.6%, p<0.001. Conclusions: A bladder filling protocol asking patients to drink less before treatment was successfully implemented in our center. We were able to eliminate incontinence during treatment and increase patient comfort. The smaller bladders lead to an increase in dose to the bladder wall, however IMRT optimization was able to produce plans that spared the bladder and were able to pass published protocol guidelines.