T Madden

Including robustness in multi-criteria optimization for intensity-modulated proton therapy

We present a method to include robustness in a multi-criteria optimization (MCO) framework for intensity-modulated proton therapy (IMPT). The approach allows one to simultaneously explore the trade-off between different objectives as well as the trade-off between robustness and nominal plan quality. In MCO, a database of plans each emphasizing different treatment planning objectives, is pre-computed to approximate the Pareto surface. An IMPT treatment plan that strikes the best balance between the different objectives can be selected by navigating on the Pareto surface. In our approach, robustness is integrated into MCO by adding robustified objectives and constraints to the MCO problem. Uncertainties (or errors) of the robust problem are modeled by pre-calculated dose-influence matrices for a nominal scenario and a number of pre-defined error scenarios (shifted patient positions, proton beam undershoot and overshoot). Objectives and constraints can be defined for the nominal scenario, thus characterizing nominal plan quality. A robustified objective represents the worst objective function value that can be realized for any of the error scenarios and thus provides a measure of plan robustness. The optimization method is based on a linear projection solver and is capable of handling large problem sizes resulting from a fine dose grid resolution, many scenarios, and a large number of proton pencil beams. A base-of-skull case is used to demonstrate the robust optimization method. It is demonstrated that the robust optimization method reduces the sensitivity of the treatment plan to setup and range errors to a degree that is not achieved by a safety margin approach. A chordoma case is analyzed in more detail to demonstrate the involved trade-offs between target underdose and brainstem sparing as well as robustness and nominal plan quality. The latter illustrates the advantage of MCO in the context of robust planning. For all cases examined, the robust optimization for each Pareto optimal plan takes less than 5 min on a standard computer, making a computationally friendly interface possible to the planner. In conclusion, the uncertainty pertinent to the IMPT procedure can be reduced during treatment planning by optimizing plans that emphasize different treatment objectives, including robustness, and then interactively seeking for a most-preferred one from the solution Pareto surface.

Dose Uncertainties in IMPT for Oropharyngeal Cancer in the Presence of Anatomical, Range, and Setup Errors

PURPOSE Setup, range, and anatomical uncertainties influence the dose delivered with intensity modulated proton therapy (IMPT), but clinical quantification of these errors for oropharyngeal cancer is lacking. We quantified these factors and investigated treatment fidelity, that is, robustness, as influenced by adaptive planning and by applying more beam directions. METHODS AND MATERIALS We used an in-house treatment planning system with multicriteria optimization of pencil beam energies, directions, and weights to create treatment plans for 3-, 5-, and 7-beam directions for 10 oropharyngeal cancer patients. The dose prescription was a simultaneously integrated boost scheme, prescribing 66 Gy to primary tumor and positive neck levels (clinical target volume-66 Gy; CTV-66 Gy) and 54 Gy to elective neck levels (CTV-54 Gy). Doses were recalculated in 3700 simulations of setup, range, and anatomical uncertainties. Repeat computed tomography (CT) scans were used to evaluate an adaptive planning strategy using nonrigid registration for dose accumulation. RESULTS For the recalculated 3-beam plans including all treatment uncertainty sources, only 69% (CTV-66 Gy) and 88% (CTV-54 Gy) of the simulations had a dose received by 98% of the target volume (D98%) >95% of the prescription dose. Doses to organs at risk (OARs) showed considerable spread around planned values. Causes for major deviations were mixed. Adaptive planning based on repeat imaging positively affected dose delivery accuracy: in the presence of the other errors, percentages of treatments with D98% >95% increased to 96% (CTV-66 Gy) and 100% (CTV-54 Gy). Plans with more beam directions were not more robust. CONCLUSIONS For oropharyngeal cancer patients, treatment uncertainties can result in significant differences between planned and delivered IMPT doses. Given the mixed causes for major deviations, we advise repeat diagnostic CT scans during treatment, recalculation of the dose, and if required, adaptive planning to improve adequate IMPT dose delivery.

Anatomic feature-based registration for patient set-up in head and neck cancer radiotherapy

Modern radiotherapy equipment is capable of delivering high precision conformal dose distributions relative to isocentre. One of the barriers to precise treatments is accurate patient re-positioning before each fraction of treatment. At Massachusetts General Hospital, we perform daily patient alignment using radiographs, which are captured by flat panel imaging devices and sent to an analysis program. A trained therapist manually selects anatomically significant features in the skeleton, and couch movement is computed based on the image coordinates of the features. The current procedure takes about 5 to 10 min and significantly affects the efficiency requirement in a busy clinic. This work presents our effort to develop an improved, semi-automatic procedure that uses the manually selected features from the first treatment fraction to automatically locate the same features on the second and subsequent fractions. An implementation of this semi-automatic procedure is currently in clinical use for head and neck tumour sites. Radiographs collected from 510 patient set-ups were used to test this algorithm. A mean difference of 1.5 mm between manual and automatic localization of individual features and a mean difference of 0.8 mm for overall set-up were seen.

Impact of Spot Size and Beam-Shaping Devices on the Treatment Plan Quality for Pencil Beam Scanning Proton Therapy

PURPOSE This study aimed to assess the clinical impact of spot size and the addition of apertures and range compensators on the treatment quality of pencil beam scanning (PBS) proton therapy and to define when PBS could improve on passive scattering proton therapy (PSPT). METHODS AND MATERIALS The patient cohort included 14 pediatric patients treated with PSPT. Six PBS plans were created and optimized for each patient using 3 spot sizes (∼12-, 5.4-, and 2.5-mm median sigma at isocenter for 90- to 230-MeV range) and adding apertures and compensators to plans with the 2 larger spots. Conformity and homogeneity indices, dose-volume histogram parameters, equivalent uniform dose (EUD), normal tissue complication probability (NTCP), and integral dose were quantified and compared with the respective PSPT plans. RESULTS The results clearly indicated that PBS with the largest spots does not necessarily offer a dosimetric or clinical advantage over PSPT. With comparable target coverage, the mean dose (Dmean) to healthy organs was on average 6.3% larger than PSPT when using this spot size. However, adding apertures to plans with large spots improved the treatment quality by decreasing the average Dmean and EUD by up to 8.6% and 3.2% of the prescribed dose, respectively. Decreasing the spot size further improved all plans, lowering the average Dmean and EUD by up to 11.6% and 10.9% compared with PSPT, respectively, and eliminated the need for beam-shaping devices. The NTCP decreased with spot size and addition of apertures, with maximum reduction of 5.4% relative to PSPT. CONCLUSIONS The added benefit of using PBS strongly depends on the delivery configurations. Facilities limited to large spot sizes (>∼8 mm median sigma at isocenter) are recommended to use apertures to reduce treatment-related toxicities, at least for complex and/or small tumors.

Effects of spot parameters in pencil beam scanning treatment planning

BACKGROUND Spot size σ (in air at isocenter), interspot spacing d, and spot charge q influence dose delivery efficiency and plan quality in Intensity Modulated Proton Therapy (IMPT) treatment planning. The choice and range of parameters varies among different manufacturers. The goal of this work is to demonstrate the influence of the spot parameters on dose quality and delivery in IMPT treatment plans, to show their interdependence, and to make practitioners aware of the spot parameter values for a certain facility. Our study could help as a guideline to make the trade-off between treatment quality and time in existing PBS centers and in future systems. METHODS We created plans for seven patients and a phantom, with different tumor sites and volumes, and compared the effect of small-, medium-, and large-spot widths (σ = 2.5, 5, and 10 mm) and interspot distances (1σ, 1.5σ, and 1.75σ) on dose, spot charge, and treatment time. Moreover, we quantified how postplanning charge threshold cuts affect plan quality and the total number of spots to deliver, for different spot widths and interspot distances. We show the effect of a minimum charge (or MU) cutoff value for a given proton delivery system. RESULTS Spot size had a strong influence on dose: larger spots resulted in more protons delivered outside the target region. We observed dose differences of 2-13 Gy (RBE) between 2.5 mm and 10 mm spots, where the amount of extra dose was due to dose penumbra around the target region. Interspot distance had little influence on dose quality for our patient group. Both parameters strongly influence spot charge in the plans and thus the possible impact of postplanning charge threshold cuts. If such charge thresholds are not included in the treatment planning system (TPS), it is important that the practitioner validates that a given combination of lower charge threshold, interspot spacing, and spot size does not result in a plan degradation. Low average spot charge occurs for small spots, small interspot distances, many beam directions, and low fractional dose values. CONCLUSIONS The choice of spot parameters values is a trade-off between accelerator and beam line design, plan quality, and treatment efficiency. We recommend the use of small spot sizes for better organ-at-risk sparing and lateral interspot distances of 1.5σ to avoid long treatment times. We note that plan quality is influenced by the charge cutoff. Our results show that the charge cutoff can be sufficiently large (i.e., 106 protons) to accommodate limitations on beam delivery systems. It is, therefore, not necessary per se to include the charge cutoff in the treatment planning optimization such that Pareto navigation (e.g., as practiced at our institution) is not excluded and optimal plans can be obtained without, perhaps, a bias from the charge cutoff. We recommend that the impact of a minimum charge cut impact is carefully verified for the spot sizes and spot distances applied or that it is accommodated in the TPS.

Impact of spot charge inaccuracies in IMPT treatments

Background: Spot charge is one parameter of pencil‐beam scanning dose delivery system whose accuracy is typically high but whose required value has not been investigated. In this work we quantify the dose impact of spot charge inaccuracies on the dose distribution in patients. Knowing the effect of charge errors is relevant for conventional proton machines, as well as for new generation proton machines, where ensuring accurate charge may be challenging. Methods: Through perturbation of spot charge in treatment plans for seven patients and a phantom, we evaluated the dose impact of absolute (up to 5× 106 protons) and relative (up to 30%) charge errors. We investigated the dependence on beam width by studying scenarios with small, medium and large beam sizes. Treatment plan statistics included the Γ passing rate, dose‐volume‐histograms and dose differences. Results: The allowable absolute charge error for small spot plans was about 2× 106 protons. Larger limits would be allowed if larger spots were used. For relative errors, the maximum allowable error size for small, medium and large spots was about 13%, 8% and 6% for small, medium and large spots, respectively. Conclusions: Dose distributions turned out to be surprisingly robust against random spot charge perturbation. Our study suggests that ensuring spot charge errors as small as 1–2% as is commonly aimed at in conventional proton therapy machines, is clinically not strictly needed.

SU-G-TeP4-04: An Automated Monte Carlo Based QA Framework for Pencil Beam Scanning Treatments

PURPOSE Prior to treating new PBS field, multiple (three) patient-field-specific QA measurements are performed: two 2D dose distributions at shallow depth (M1) and at the tumor depth (M2) with treatment hardware at zero gantry angle; one 2D dose distribution at iso-center (M3) without patient specific devices at the planned gantry angle. This patient-specific QA could be simplified by the use of MC model. The results of MC model commissioning for a spot-scanning system and the fully automated TOPAS/MC-based QA framework will be presented. METHODS We have developed in-house MC interface to access a TPS (Astroid) database from a computer cluster remotely. Once a plan is identified, the interface downloads information for the MC simulations, such as patient images, apertures points, and fluence maps and initiates calculations in both the patient and QA geometries. The resulting calculations are further analyzed to evaluate the TPS dose accuracy and the PBS delivery. RESULTS The Monte Carlo model of our system was validated within 2.0 % accuracy over the whole range of the dose distribution (proximal/shallow part, as well as target dose part) due to the location of the measurements. The averaged range difference after commissioning was 0.25 mm over entire treatment ranges, e.g., 6.5 cm to 31.6 cm. CONCLUSION As M1 depths range typically from 1 cm to 4 cm from the phantom surface, The Monte Carlo model of our system was validated within +- 2.0 % in absolute dose level over a whole treatment range. The averaged range difference after commissioning was 0.25 mm over entire treatment ranges, e.g., 6.5 cm to 31.6 cm. This work was supported by NIH/NCI under CA U19 21239.

SU‐E‐T‐512: Monte Carlo Dose Verification of Pencil Beam Scanning Proton Therapy

Purpose: To verify a clinical pencil (PB) beam dose calculation algorithm for scanned beam intensity modulated proton therapy (IMPT), using TOPAS (TOol for PArticle Simulation), a GEANT4 based Monte Carlo (MC) simulation system. Methods: Seven patients, previously treated with IMPT for various treatment sites and prescriptions, were selected from our patient database. Proton fluence maps of the treated plans were exported for each field from our clinical treatment planning system (ASTROID) and imported to TOPAS along with the patient and beam geometry. The absolute dose distribution of each individual beam was calculated and compared to the PB algorithm‐based calculation from ASTROID. Results: The differences observed in mean and median target doses were less than ±1% for all cases, while D02 and D98 (surrogates for maximum and minimum dose values respectively) differed by less than ±3% for the majority of beams. Differences in the mean dose for the organs at risk (OARs) ranged from −8.9% to 3.7%, with reference to MC calculation, with an average over all the OARs of −0.1%, indicating no systematic over‐or under‐estimation of the dose by the PB algorithm. 3D gamma analysis (2%/2mm) for the PB to MC dose comparison resulted in an average 95.2% (±5.0) of the target volume having an absolute gamma value equal or less than 1 and 99.2% (±1.2%) equal or less than 2. For the healthy tissue receiving at least 5% of the target mean dose, the corresponding percentages were 99.6% (±0.3%) and 99.9% (±0.1%). Conclusion: We have clinically implemented MC for IMPT plan recalculation. Our PB calculation algorithm for IMPT was found to be in overall good agreement with MC calculations. Clinically significant deviations in OAR mean dose can be attributed to lung tissue or bone anatomy in the beam path.

TH‐A‐213AB‐08: Robust Multi‐Criteria IMPT Optimization

Purpose: We present a method to include robustness in a multi‐criteria optimization (MCO) framework for intensity‐modulated proton therapy (IMPT). The approach allows one to simultaneously explore the trade‐off between different objectives including robustness and nominal plan quality. Methods: A database of plans each emphasizing different treatment planning objectives, is pre‐computed to approximate the Pareto surface. An IMPT treatment plan that strikes the best balance between the different objectives can be selected by navigating on the Pareto surface. We integrate robustness into MCO by adding robustified objectives and constraints. Uncertainties are modeled by pre‐calculated dose‐influence matrices for a nominal scenario and a number of pre‐defined error scenarios (shifted patient positions, proton beam undershoot and overshoot). A robustified objective represents the worst objective function value that can be realized for any of the error scenarios and thus provides a measure of plan robustness. The optimization method uses a linear projection solver and is capable of handling large problem sizes resulting from a fine dose grid resolution, many scenarios, and a large number of proton pencil beams. Results: A base‐of‐skull case is used to demonstrate that the robust optimization method reduces the sensitivity of the treatment plan to setup and range errors to a degree that is not achieved by a safety margin approach. A chordoma case is analyzed in more detail to demonstrate the involved trade‐offs between target underdose and brainstem sparing as well as robustness and nominal plan quality. The robust optimization for each Pareto optimal plan takes less than 5 min on a standard computer. Conclusions: The uncertainty pertinent to the IMPT procedure can be reduced during treatment planning by optimizing a database of plans that emphasize different treatment objectives, including robustness. The planner can then interactively explore all convex combinations of database plans to decide on the most‐preferred trade‐off.

TH‐A‐BRA‐02: Dose Uncertainties in IMPT for Oropharyngeal Cancer in the Presence of Anatomical, Setup and Range Errors

Purpose: Setup, range and anatomical uncertainties influence the dose delivered with proton pencil‐beams, but a quantification of these errors for intensity‐modulated proton therapy (IMPT) for oropharyngeal cancer is lacking. The purpose of this work is to quantify these effects. We also investigate whether treatment plans are more robust against errors when more beam directions are applied. Methods: We used an in‐house developed inverse treatment planning system for proton pencil‐beam scanning, which performs multi‐criteria optimization. First, treatment plans for 3, 5, and 7 beam directions were created for 5 oropharyngeal cancer patients. A simultaneous‐integrated boost technique was used, prescribing 66Gy and 54Gy to high and low dose regions, respectively, delivered in 30 fractions. Second, 300 treatment simulations were performed, recalculating the dose while including uncertainties. Anatomical uncertainties were taken into account by using two CT scans per patient. For setup errors an online setup‐protocol was simulated. DVH parameters were used for dose evaluation. Results: The treatment plans were of high quality, with good target coverage and excellent OAR sparing. However, setup, range and anatomical uncertainties can lead to large differences between the planned and delivered dose. For the 3‐beam plans, the expected V95% of the high‐ dose CTV reduced from 100% (planned) to on average 92%(range:82–98%), and the expected V107% increased from 1.3% (planned) to on average 4.2%(range:0.1%‐6.7%). The expected V95% of the low‐dose CTV reduced from 100% (planned) to on average 94% (range:86‐99%). The dose to the OARs generally increased. The plans with more beam directions were not more robust against errors (p>0.05). Conclusions: Setup, range and anatomical uncertainties in IMPT for oropharyngeal cancer patients can lead to considerable underdosage of the target volume, which should be accounted for by replanning and/or robust optimization techniques. Robustness is not achieved by increasing the number of beam directions included in the treatment plan.