Monitor unit verification for Varian TrueBeam VMAT plans using Monte Carlo calculations and phase space data.

To use the open-source Monte Carlo (MC) software calculations for TPS monitor unit verification of VMAT plans, delivered with the Varian TrueBeam linear accelerator, and compare the results with a commercial software product, following the guidelines set in AAPM Task Group 219. The TrueBeam is modeled in EGSnrc using the Varian-provided phase-space files. Thirteen VMAT TrueBeam treatment plans representing various anatomical regions were evaluated, comprising 37 treatment arcs. VMAT plans simulations were performed on a computing cluster, using 107 -109 particle histories per arc. Point dose differences at five reference points per arc were compared between Eclipse, MC, and the commercial software, MUCheck. MC simulation with 5 × 107 histories per arc offered good agreement with Eclipse and a reasonable average calculation time of 9-18 min per full plan. The average absolute difference was 3.0%, with only 22% of all points exceeding the 5% action limit. In contrast, the MUCheck average absolute difference was 8.4%, with 60% of points exceeding the 5% dose difference. Lung plans were particularly problematic for MUCheck, with an average absolute difference of approximately 16%. Our EGSnrc-based MC framework can be used for the MU verification of VMAT plans calculated for the Varian TrueBeam; furthermore, our phase space approach can be adapted to other treatment devices by using appropriate phase space files. The use of 5 × 107 histories consistently satisfied the 5% action limit across all plan types for the majority of points, performing significantly better than a commercial MU verification system, MUCheck. As faster processors and cloud computing facilities become even more widely available, this approach can be readily implemented in clinical settings.

A Hamiltonian Engine for Radiotherapy Optimization.

We apply a new hardware and software platform called the Hamiltonian Engine for Radiotherapy Optimization (HERO) to the problem of Intensity-Modulated Radiation Therapy (IMRT) treatment planning. HERO solves large general-form binary optimization problems by decomposing them into sub-problems and approximating them using a quadratic pseudo-boolean function. Optimizing the resulting function becomes a quadratic unconstrained binary optimization (QUBO) problem, which has been widely studied and has numerous applications in various fields. A Quantum Annealer (QA) approach has been previously investigated to solve QUBO problems, including IMRT optimization. However, the QA can only accommodate a small number of variables and requires several hours to obtain optimized plans. HERO acts as an optimizer for QUBO problems, which not only addresses these shortcomings but also relies solely on conventional hardware design while operating at room temperature. We evaluate HERO on seven prostate IMRT cases with clinical objectives, each using approximately 6000 beamlets. Our method was compared to the commercial treatment planning software, Eclipse, for both time-to-solution and plan quality. HERO solves most cases in about 30 seconds, with significantly lower objective function scores than Eclipse. The results indicate that HERO is promising for radiation therapy optimization problems. Additionally, HERO has the potential to be applied to Volumetric-Modulated Arc Therapy (VMAT) and other complex types of treatment planning.

Enhanced optimization of volumetric modulated arc therapy plans using Monte Carlo generated beamlets.

PURPOSE: A treatment planning system (TPS) produces volumetric modulated arc therapy (VMAT) plans by applying an optimization process to an objective function, followed by an accurate calculation of the final, deliverable dose. However, during the optimization step, a rapid dose calculation algorithm is required, which reduces its accuracy and its representation of the objective function space. Monte Carlo (MC) routines, considered the gold standard in accuracy, are currently too slow for practical comprehensive VMAT optimization. Therefore, we propose a novel approach called enhanced optimization (EO), which employs the TPS VMAT plan as a starting point, and applies small perturbations to nudge the solution closer to a true objective minimum. The perturbations consist of beamlet dose matrices, calculated using MC routines on a distributed-computing framework. METHODS: DICOM files for clinical VMAT plans files are exported from the TPS and used to generate input files for the EGSnrc MC toolkit. Beamlet doses are calculated using the MC routines, each corresponding to a single multileaf collimator leaf from a single control point traveling 0.5 cm in or out of the field. A typical VMAT plan requires 5000 to 10 000 beamlets, which may be calculated overnight. This results in a ternary-valued objective function, which may use the same clinical objectives as the original VMAT plan. A simple greedy search algorithm is applied to minimize this function and determine the optimal set of ternary variables. The resulting modified control point parameters are imported into the TPS to calculate the final, deliverable dose, and to compare the EO plan with the original. EO was evaluated retrospectively on seven VMAT plans (two adult brain, one pediatric brain, two head and neck, and two prostate). Additionally, the use of stricter objectives was investigated for two of the cases: the left cochlea planning organ at risk (OAR) volume objective for the pediatric brain case, and the rectum objective for a prostate case. RESULTS: EO produced improved objective scores (by 6% to 60%) and dose-volume histograms (DVH) for the brain plans and the head and neck plans. For each of these plans, the target dose minimum and homogeneity were preserved, while one or more of the OAR DVH's was reduced. Although EO also reduced the objective scores for the prostate plans (by 46% and 79%), their absolute score and DVH improvements were not substantial. The stricter objective on the pediatric brain case resulted in lower dose to the OAR without compromising the target dose. However, the rectum dose in the prostate case could not be improved without reducing dose homogeneity to the planning target volume, suggesting that VMAT prostate cases may already be highly optimized by the TPS. CONCLUSION: We have developed a novel approach to improving the dose distribution of VMAT plans, which relies on MC calculations to provide small modifications to the control points. This method may be particularly useful for complex treatments in which a certain OAR is of concern and it is difficult for the treatment planner to obtain an acceptable solution with the TPS. Further development will reduce the beamlet computation time and result in more sophisticated EO treatment planning methods.

Secondary monitor unit calculations for VMAT using parallelized Monte Carlo simulations.

We have developed a fast and accurate in-house Monte Carlo (MC) secondary monitor unit (MU) check method, based on the EGSnrc system, for independent verification of volumetric modulated arc therapy (VMAT) treatment planning system dose calculations, in accordance with TG-114 recommendations. For a VMAT treatment plan created for a Varian Trilogy linac, DICOM information was exported from Eclipse. An open-source platform was used to generate input files for dose calculations using the EGSnrc framework. The full VMAT plan simulation employed 107 histories, and was parallelized to run on a computer cluster. The resulting 3ddose matrices were converted to the DICOM format using CERR and imported into Eclipse. The method was evaluated using 35 clinical VMAT plans of various treatment sites. For each plan, the doses calculated with the MC approach at four three-dimensional reference points were compared to the corresponding Eclipse calculations, as well as calculations performed using the clinical software package, MUCheck. Each MC arc simulation of 107 particles required 13-25 min of total time, including processing and calculation. The average discrepancies in calculated dose values between the MC method and Eclipse were 2.03% (compared to 3.43% for MUCheck) for prostate cases, 2.45% (3.22% for MUCheck) for head and neck cases, 1.7% (5.51% for MUCheck) for brain cases, and 2.84% (5.64% for MUCheck) for miscellaneous cases. Of 276 comparisons, 201 showed greater agreement between the treatment planning system and MC vs MUCheck. The largest discrepancies between MC and MUCheck were found in regions of high dose gradients and heterogeneous densities. By parallelizing the calculations, point-dose accuracies of 2-7%, sufficient for clinical secondary checks, can be achieved in a reasonable amount of time. As computer clusters and/or cloud computing become more widespread, this method will be useful in most clinical setups.

Adapting VMAT plans optimized for an HD120 MLC for delivery with a Millennium MLC.

Linac downtime invariably impacts delivery of patients' scheduled treatments. Transferring a patient's treatment to an available linac is a common practice. Transferring a Volumetric Modulated Arc Therapy (VMAT) plan from a linac equipped with a standard-definition MLC to one equipped with a higher definition MLC is practical and routine in clinics with multiple MLC-equipped linacs. However, the reverse transfer presents a challenge because the high-definition MLC aperture shapes must be adapted for delivery with the lower definition device. We have developed an efficient method to adapt VMAT plans originally designed for a high-definition MLC to a standard-definition MLC. We present the dosimetric results of our adaptation method for head-and-neck, brain, lung, and prostate VMAT plans. The delivery of the adapted plans was verified using standard phantom measurements.

Comparison of hybrid volumetric modulated arc therapy (VMAT) technique and double arc VMAT technique in the treatment of prostate cancer.

BACKGROUND: Volumetric modulated arc therapy (VMAT) has quickly become accepted as standard of care for the treatment of prostate cancer based on studies showing it is able to provide faster delivery with adequate target coverage and reduced monitor units while maintaining organ at risk (OAR) sparing. This study aims to demonstrate the potential to increase dose conformality with increased planner control and OAR sparing using a hybrid treatment technique compared to VMAT. METHODS: Eleven patients having been previously treated for prostate cancer with VMAT techniques were replanned with a hybrid technique on Varian Treatment Planning System. Multiple static IMRT fields (2 to 3) were planned initially based on critical OAR to reduce dose but provide some planning treatment volume (PTV) coverage. This was used as a base dose plan to provide 30-35% coverage for a single arc VMAT plan. RESULTS: The clinical VMAT plan was used as a control for the purposes of comparison. Average of all OAR sparing between the hybrid technique and VMAT showed the hybrid plan delivering less dose in almost all cases except for V80 of the bladder and maximum dose to right femoral head. PTV coverage was superior with the VMAT technique. Monitor unit differences varied, with the hybrid plan able to deliver fewer units 37% of the time, similar results 18% of the time, and higher units 45% of the time. On average, the hybrid plan delivered 10% more monitor units. CONCLUSIONS: The hybrid plan can be delivered in a single gantry rotation combining aspects of VMAT with regions of dynamic intensity modulated radiation therapy (IMRT) within the treatment arc.

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.

Quantification of incidental mediastinal and hilar irradiation delivered during definitive stereotactic body radiation therapy for peripheral non-small cell lung cancer.

To determine the amount of incidental radiation dose received by the mediastinal and hilar nodes for patients with non-small cell lung cancer (NSCLC) treated with stereotactic body radiation therapy (SBRT). Fifty consecutive patients with NSCLC, treated using an SBRT technique, were identified. Of these patients, 38 had a prescription dose of 60 Gy in 20-Gy fractions and were eligible for analysis. For each patient, ipsilateral upper (level 2) and lower (level 4) paratracheal, and hilar (level 10) nodal regions were contoured on the planning computed tomography (CT) images. Using the clinical treatment plan, dose and volume calculations were performed retrospectively for each nodal region. SBRT to upper lobe tumors resulted in an average total ipsilateral mean dose of between 5.2 and 7.8 Gy for the most proximal paratracheal nodal stations (2R and 4R for right upper lobe lesions, 2L and 4L for left upper lobe lesions). SBRT to lower lobe tumors resulted in an average total ipsilateral mean dose of between 15.6 and 21.5 Gy for the most proximal hilar nodal stations (10R for right lower lobe lesions, 10 l for left lower lobe lesions). Doses to more distal nodes were substantially lower than 5 Gy. The often substantial incidental irradiation, delivered during SBRT for peripheral NSCLC of the lower lobes to the most proximal hilar lymph nodes may be therapeutic for low-volume, subclinical nodal disease. Treatment of peripheral upper lobe lung tumors delivers less incidental irradiation to the paratracheal lymph nodes with lower likelihood of therapeutic benefit.

Optimization of beam angles for intensity modulated radiation therapy treatment planning using genetic algorithm on a distributed computing platform.

Planning intensity modulated radiation therapy (IMRT) treatment involves selection of several angle parameters as well as specification of structures and constraints employed in the optimization process. Including these parameters in the combinatorial search space vastly increases the computational burden, and therefore the parameter selection is normally performed manually by a clinician, based on clinical experience. We have investigated the use of a genetic algorithm (GA) and distributed-computing platform to optimize the gantry angle parameters and provide insight into additional structures, which may be necessary, in the dose optimization process to produce optimal IMRT treatment plans. For an IMRT prostate patient, we produced the first generation of 40 samples, each of five gantry angles, by selecting from a uniform random distribution, subject to certain adjacency and opposition constraints. Dose optimization was performed by distributing the 40-plan workload over several machines running a commercial treatment planning system. A score was assigned to each resulting plan, based on how well it satisfied clinically-relevant constraints. The second generation of 40 samples was produced by combining the highest-scoring samples using techniques of crossover and mutation. The process was repeated until the sixth generation, and the results compared with a clinical (equally-spaced) gantry angle configuration. In the sixth generation, 34 of the 40 GA samples achieved better scores than the clinical plan, with the best plan showing an improvement of 84%. Moreover, the resulting configuration of beam angles tended to cluster toward the patient's sides, indicating where the inclusion of additional structures in the dose optimization process may avoid dose hot spots. Additional parameter selection in IMRT leads to a large-scale computational problem. We have demonstrated that the GA combined with a distributed-computing platform can be applied to optimize gantry angle selection within a reasonable amount of time.

A nested partitions framework for beam angle optimization in intensity-modulated radiation therapy.

Coupling beam angle optimization with dose optimization in intensity-modulated radiation therapy (IMRT) increases the size and complexity of an already large-scale combinatorial optimization problem. We have developed a novel algorithm, nested partitions (NP), that is capable of finding suitable beam angle sets by guiding the dose optimization process. NP is a metaheuristic that is flexible enough to guide the search of a heuristic or deterministic dose optimization algorithm. The NP method adaptively samples from the entire feasible region, or search space, and coordinates the sampling effort with a systematic partitioning of the feasible region at successive iterations, concentrating the search in promising subsets. We used a 'warm-start' approach by initiating NP with beam angle samples derived from an integer programming (IP) model. In this study, we describe our implementation of the NP framework with a commercial optimization algorithm. We compared the NP framework with equi-spaced beam angle selection, the IP method, greedy heuristic and random sampling heuristic methods. The results of the NP approach were evaluated using two clinical cases (head and neck and whole pelvis) involving the primary tumor and nodal volumes. Our results show that NP produces better quality solutions than the alternative considered methods.

The use of gated and 4D CT imaging in planning for stereotactic body radiation therapy.

The localization of treatment targets is of utmost importance for patients receiving stereotactic body radiation therapy (SBRT), where the dose per fraction is large. While both setup or respiration-induced motion components affect the localization of the treatment volume, the purpose of this work is to describe our management of the intrafraction localization uncertainty induced by normal respiration. At our institution, we have implemented gated computed tomography (CT) acquisition with an active breathing control system (ABC), and 4-dimensional (4D) CT using a skin-based marker and retrospective respiration phase-based image sorting. During gated simulation, 3D CT images were acquired corresponding to end-inhalation and end-exhalation. For 4D CT imaging, 3D CT images were acquired corresponding to 8 phases of the respiratory cycle. In addition to gated or 4D CT images, we acquired a conventional free-breathing CT (FB). For both gated and 4D CT images, the target contours were registered to the FB scan in the planning system. These contours were then combined in the FB image set to form the internal target volume (ITV). Dynamic conformal arc treatment plans were generated for the ITV using the FB scan and the gated or 4D scans with an additional 7-mm margin for patient setup uncertainty. We have described our results for a pancreas and a lung tumor case. Plans were normalized so that the PTV received 95% of the prescription dose. The dose distribution for all the critical structures in the pancreas and lung tumor cases resulted in increased sparing when the ITV was defined using gated or 4D CT images than when the FB scan was used. Our results show that patient-specific target definition using gated or 4D CT scans lead to improved normal tissue sparing.

A multiplan treatment-planning framework: a paradigm shift for intensity-modulated radiotherapy.

PURPOSE: To describe a multiplan intensity-modulated radiotherapy (IMRT) planning framework, and to describe a decision support system (DSS) for ranking multiple plans and modeling the planning surface. METHODS AND MATERIALS: One hundred twenty-five plans were generated sequentially for a head-and-neck case and a pelvic case by varying the dose-volume constraints on each of the organs at risk (OARs). A DSS was used to rank plans according to dose-volume histogram (DVH) values, as well as equivalent uniform dose (EUD) values. Two methods for ranking treatment plans were evaluated: composite criteria and pre-emptive selection. The planning surface determined by the results was modeled using quadratic functions. RESULTS: The DSS provided an easy-to-use interface for the comparison of multiple plan features. Plan ranking resulted in the identification of one to three "optimal" plans. The planning surface models had good predictive capability with respect to both DVH values and EUD values and generally, errors of <6%. Models generated by minimizing the maximum relative error had significantly lower relative errors than models obtained by minimizing the sum of squared errors. Using the quadratic model, plan properties for one OAR were determined as a function of the other OAR constraint settings. The modeled plan surface can then be used to understand the interdependence of competing planning objectives. CONCLUSION: The DSS can be used to aid the planner in the selection of the most desirable plan. The collection of quadratic models constructed from the plan data to predict DVH and EUD values generally showed excellent agreement with the actual plan values.

Effects of point configuration on the accuracy in 3D reconstruction from biplane images.

Two or more angiograms are being used frequently in medical imaging to reconstruct locations in three-dimensional (3D) space, e.g., for reconstruction of 3D vascular trees, implanted electrodes, or patient positioning. A number of techniques have been proposed for this task. In this simulation study, we investigate the effect of the shape of the configuration of the points in 3D (the "cloud" of points) on reconstruction errors for one of these techniques developed in our laboratory. Five types of configurations (a ball, an elongated ellipsoid (cigar), flattened ball (pancake), flattened cigar, and a flattened ball with a single distant point) are used in the evaluations. For each shape, 100 random configurations were generated, with point coordinates chosen from Gaussian distributions having a covariance matrix corresponding to the desired shape. The 3D data were projected into the image planes using a known imaging geometry. Gaussian distributed errors were introduced in the x and y coordinates of these projected points. Gaussian distributed errors were also introduced into the gantry information used to calculate the initial imaging geometry. The imaging geometries and 3D positions were iteratively refined using the enhanced-Metz-Fencil technique. The image data were also used to evaluate the feasible R-t solution volume. The 3D errors between the calculated and true positions were determined. The effects of the shape of the configuration, the number of points, the initial geometry error, and the input image error were evaluated. The results for the number of points, initial geometry error, and image error are in agreement with previously reported results, i.e., increasing the number of points and reducing initial geometry and/or image error, improves the accuracy of the reconstructed data. The shape of the 3D configuration of points also affects the error of reconstructed 3D configuration; specifically, errors decrease as the "volume" of the 3D configuration increases, as would be intuitively expected, and shapes with larger spread, such as spherical shapes, yield more accurate reconstructions. These results are in agreement with an analysis of the solution volume of feasible geometries and could be used to guide selection of points for reconstruction of 3D configurations from two views.

Vessel size measurements in angiograms: manual measurements.

Vessel size measurement is perhaps the most often performed quantitative analysis in diagnostic and interventional angiography. Although automated vessel sizing techniques are generally considered to have good accuracy and precision, we have observed that clinicians rarely use these techniques in standard clinical practice, choosing to indicate the edges of vessels and catheters to determine sizes and calibrate magnifications, i.e., manual measurements. Thus, we undertook an investigation of the accuracy and precision of vessel sizes calculated from manually indicated edges of vessels. Manual measurements were performed by three neuroradiologists and three physicists. Vessel sizes ranged from 0.1-3.0 mm in simulation studies and 0.3-6.4 mm in phantom studies. Simulation resolution functions had full-widths-at-half-maximum (FWHM) ranging from 0.0 to 0.5 mm. Phantom studies were performed with 4.5 in., 6 in., 9 in., and 12 in. image intensifier modes, magnification factor = 1, with and without zooming. The accuracy and reproducibility of the measurements ranged from 0.1 to 0.2 mm, depending on vessel size, resolution, and pixel size, and zoom. These results indicate that manual measurements may have accuracies comparable to automated techniques for vessels with sizes greater than 1 mm, but that automated techniques which take into account the resolution function should be used for vessels with sizes smaller than 1 mm.

Vessel size measurements in angiograms: a comparison of techniques.

As interventional procedures become more complicated, the need for accurate quantitative vascular information increases. In response to this need, many commercial vendors provide techniques for measurement of vessel sizes, usually based on derivative techniques. In this study, we investigate the accuracy of several techniques used in the measurement of vessel size. Simulated images of vessels having circular cross sections were generated and convolved with various focal spot distributions taking into account the magnification. These vessel images were then convolved with Gaussian image detector line spread functions (LSFs). Additionally, images of a phantom containing vessels with a range of diameters were acquired for the 4.5", 6", 9", and 12" modes of an image intensifier-TV (II-TV) system. Vessel sizes in the images were determined using a first-derivative technique, a second-derivative technique, a linear combination of these two measured sizes, a thresholding technique, a densitometric technique, and a model-based technique. For the same focal spot size, the shape of the focal spot distribution does not affect measured vessel sizes except at large magnifications. For vessels with diameters larger than the full-width-at-half-maximum (FWHM) of the LSF, accurate vessel sizes (errors approximately 0.1 mm) could be obtained by using an average of sizes determined by the first and second derivatives. For vessels with diameters smaller than the FWHM of the LSF, the densitometric and model-based techniques can provide accurate vessel sizes when these techniques are properly calibrated.