E Ramirez

Commissioning of output factors for uniform scanning proton beams

PURPOSE Current commercial treatment planning systems are not able to accurately predict output factors and calculate monitor units for proton fields. Patient-specific field output factors are thus determined by either measurements or empirical modeling based on commissioning data. The objective of this study is to commission output factors for uniform scanning beams utilized at the ProCure proton therapy centers. METHODS Using water phantoms and a plane parallel ionization chamber, the authors first measured output factors with a fixed 10 cm diameter aperture as a function of proton range and modulation width for clinically available proton beams with ranges between 4 and 31.5 cm and modulation widths between 2 and 15 cm. The authors then measured the output factor as a function of collimated field size at various calibration depths for proton beams of various ranges and modulation widths. The authors further examined the dependence of the output factor on the scanning area (i.e., uncollimated proton field), snout position, and phantom material. An empirical model was developed to calculate the output factor for patient-specific fields and the model-predicted output factors were compared to measurements. RESULTS The output factor increased with proton range and field size, and decreased with modulation width. The scanning area and snout position have a small but non-negligible effect on the output factors. The predicted output factors based on the empirical modeling agreed within 2% of measurements for all prostate treatment fields and within 3% for 98.5% of all treatment fields. CONCLUSIONS Comprehensive measurements at a large subset of available beam conditions are needed to commission output factors for proton therapy beams. The empirical modeling agrees well with the measured output factor data. This investigation indicates that it is possible to accurately predict output factors and thus eliminate or reduce time-consuming patient-specific output measurements for proton treatments.

Proton therapy vs. VMAT for prostate cancer: a treatment planning study

Abstract Purpose: The main objective of this study was to compare the dosimetric quality of volumetric modulated arc therapy (VMAT) with that of proton therapy for high-risk prostate cancer. Patients and Materials: Twelve patients with high-risk prostate cancer previously treated with uniform scanning proton therapy (USPT) were included in this study. Proton planning was done using the XiO treatment planning system (TPS) with two 1800 parallel-opposed lateral fields. The VMAT planning was done using the RapidArc technique with two arcs in the Eclipse TPS. The VMAT and proton plans were calculated using the anisotropic analytical algorithm and pencil-beam algorithm, respectively. The calculated VMAT and proton plans were then normalized so that at least 95% of the planning target volume (PTV) received the prescription dose. The dosimetric evaluation was performed by comparing the physical dose-volume parameters, which were obtained from the VMAT and proton plans. Results: The average difference in the PTV ...

Measurements of lateral penumbra for uniform scanning proton beams under various beam delivery conditions and comparison to the XiO treatment planning system

PURPOSE The main purposes of this study were to (1) investigate the dependency of lateral penumbra (80%-20% distance) of uniform scanning proton beams on various factors such as air gap, proton range, modulation width, compensator thickness, and depth, and (2) compare the lateral penumbra calculated by a treatment planning system (TPS) with measurements. METHODS First, lateral penumbra was measured using solid-water phantom and radiographic films for (a) air gap, ranged from 0 to 35 cm, (b) proton range, ranged from 8 to 30 cm, (c) modulation, ranged from 2 to 10 cm, (d) compensator thickness, ranged from 0 to 20 cm, and (e) depth, ranged from 7 to 15 cm. Second, dose calculations were computed in a virtual water phantom using the XiO TPS with pencil beam algorithm for identical beam conditions and geometrical configurations that were used for the measurements. The calculated lateral penumbra was then compared with the measured one for both the horizontal and vertical scanning magnets of our uniform scanning proton beam delivery system. RESULTS The results in the current study showed that the lateral penumbra of horizontal scanning magnet was larger (up to 1.4 mm for measurement and up to 1.0 mm for TPS) compared to that of vertical scanning magnet. Both the TPS and measurements showed an almost linear increase in lateral penumbra with increasing air gap as it produced the greatest effect on lateral penumbra. Lateral penumbra was dependent on the depth and proton range. Specifically, the width of lateral penumbra was found to be always lower at shallower depth than at deeper depth within the spread out Bragg peak (SOBP) region. The lateral penumbra results were less sensitive to the variation in the thickness of compensator, whereas lateral penumbra was independent of modulation. Overall, the comparison between the results of TPS with that of measurements indicates a good agreement for lateral penumbra, with TPS predicting higher values compared to measurements. CONCLUSIONS Lateral penumbra of uniform scanning proton beams depends on air gap, proton range, compensator thickness, and depth, whereas lateral penumbra is not dependent on modulation. The XiO TPS typically overpredicted lateral penumbra compared to measurements, within 1 mm for most cases, but the difference could be up to 2.5 mm at a deep depth and large air gap.

WE-G-BRA-01: Patient Safety and Treatment Quality Improvement Through Incident Learning: Experience of a Non-Academic Proton Therapy Center

Purpose: Incident learning has been proven to improve patient safety and treatment quality in conventional radiation therapy. However, its application in proton therapy has not been reported yet to our knowledge. In this study, we report our experience in developing and implementation of an in-house incident learning system. Methods: An incident learning system was developed based on published principles and tailored for our clinical practice and available resource about 18 months ago. The system includes four layers of error detection and report: 1) dosimetry peer review; 2) physicist plan quality assurance (QA); 3) treatment delivery issue on call and record; and 4) other incident report. The first two layers of QA and report were mandatory for each treatment plan through easy-to-use spreadsheets that are only accessible by the dosimetry and physicist departments. The treatment delivery issues were recorded case by case by the on call physicist. All other incidents were reported through an online incident report system, which can be anonymous. The incident report includes near misses on planning and delivery, process deviation, machine issues, work flow and documentation. Periodic incident reviews were performed. Results: In total, about 116 errors were reported through dosimetry review, 137 errors through plan QA, 83 treatment issues through physics on call record, and 30 through the online incident report. Only 8 incidents (2.2%) were considered to have a clinical impact to patients, and the rest of errors were either detected before reaching patients or had negligible dosimetric impact (<5% dose variance). Personnel training & process improvements were implemented upon periodic incident review. Conclusion: An incident learning system can be helpful in personnel training, error reduction, and patient safety and treatment quality improvement. The system needs to be catered for each clinic’s practice and available resources. Incident and knowledge sharing among proton centers are encouraged.

SU-E-T-170: Evaluation of Rotational Errors in Proton Therapy Planning of Lung Cancer

PURPOSE To investigate the impact of rotational (roll, yaw, and pitch) errors in proton therapy planning of lung cancer. METHODS A lung cancer case treated at our center was used in this retrospective study. The original plan was generated using two proton fields (posterior-anterior and left-lateral) with XiO treatment planning system (TPS) and delivered using uniform scanning proton therapy system. First, the computed tomography (CT) set of original lung treatment plan was re-sampled for rotational (roll, yaw, and pitch) angles ranged from -5° to +5°, with an increment of 2.5°. Second, 12 new proton plans were generated in XiO using the 12 re-sampled CT datasets. The same beam conditions, isocenter, and devices were used in new treatment plans as in the original plan. All 12 new proton plans were compared with original plan for planning target volume (PTV) coverage and maximum dose to spinal cord (cord Dmax). RESULTS PTV coverage was reduced in all 12 new proton plans when compared to that of original plan. Specifically, PTV coverage was reduced by 0.03% to 1.22% for roll, by 0.05% to 1.14% for yaw, and by 0.10% to 3.22% for pitch errors. In comparison to original plan, the cord Dmax in new proton plans was reduced by 8.21% to 25.81% for +2.5° to +5° pitch, by 5.28% to 20.71% for +2.5° to +5° yaw, and by 5.28% to 14.47% for -2.5° to -5° roll. In contrast, cord Dmax was increased by 3.80% to 3.86% for -2.5° to -5° pitch, by 0.63% to 3.25% for -2.5° to -5° yaw, and by 3.75% to 4.54% for +2.5° to +5° roll. CONCLUSION PTV coverage was reduced by up to 3.22% for rotational error of 5°. The cord Dmax could increase or decrease depending on the direction of rotational error, beam angles, and the location of lung tumor.

SU‐E‐T‐190: Manufacturing Accuracy of Range Compensators — Analysis of Over 12,000 Milling Points

PURPOSE Range compensators are used in both passive and active uniform scanning proton beams to conform proton dose to the targets distal boundary. The manufacturing accuracy of compensators has a direct effect on range uncertainty in proton beams. The purpose of the study is to investigate the thickness difference between manufactured and planned compensators and its effect on range uncertainty. METHODS We used a range compensator QA table with 2-D motors and overhead 1D laser measurement tool to measure the compensator thickness. The QA device allows automatic multiple milling point measurements with minimal man-hours, with export of QA results to a tabulated file for ease of data analysis. We recorded the QA results of over 12,000 points across over 500 different blue wax compensators and analyzed the thickness deviation from treatment planning with a histogram plot. RESULTS The average measured compensator thickness is about 0.03 mm less than the expected according to the treatment plans. The standard deviation is about 0.17 mm. Causes for thickness deviation of larger than 0.5 mm were investigated. The corresponding water equivalent range uncertainty of a compensator is similar to the physical thickness deviation because the blue wax has a relative stopping power ratio of 0.97. Some large deviations due to human errors were analyzed separately. CONCLUSION The range uncertainty through the range compensator is dependent on the accuracy of the milling device and also the relative stopping power of the compensator material. It is recommended to use material with stopping power close to water for compensators to mitigate the errors in milling yet also to reduce the bulkiness of the compensator. The range uncertainty due to compensator thickness inaccuracy is estimated to be about 0.3 mm (2 sigma) in our clinic.

SU‐E‐T‐138: Range Verification for Proton Therapy Systems: A Multi‐Center Study

PURPOSE Proton beams have finite range and spare normal tissues distal to the target. Accurate determination and routine verification of proton range are critical in proton therapy, but depends on measurement device and method and could vary with proton centers. The purpose of this study is to develop a consistent approach to cross verify range accuracy at various proton centers. METHODS Three Lucite blocks were manufactured, with water equivalent thicknesses (WET) of about 7, 9 and 12 cm. The exact amount of WET was determined using a multi-layer ionization chamber (MLIC). Using various block combinations, we tested four beams of different proton ranges (9, 16, 21 and 28 cm). The block combination and proton beams were chosen so that the measured dose by the PPC was about 50% of the dose at the center of spread out Bragg peak (SOBP) for each beam. A parallel plane chamber (PPC) was used to measure the dose at the center of SOBP and immediately behind the block(s). Since proton beams have a very sharp distal falloff (∼10-20% dose change per mm), the PPC reading behind the blocks is very sensitive to range change, thus allowing accurate range verification. RESULTS A set of Lucite blocks with step-by-step instructions were developed for cross center proton range checks. Using these blocks and a commercial PPC, various ranges were measured and compared to MLIC measurement and nominal ranges. Such measurements were carried out for four treatment rooms at one center, and cross range checks at other proton centers are under way. CONCLUSION A simple method has been developed to cross check proton ranges among various proton centers within sub-millimeter accuracy. The same approach can also be used to check proton range consistency with time. The multi-center range results would provide valuable data for range uncertainty assessment.

SU‐E‐J‐33: Proton Therapy Fiducial Comparison

PURPOSE Fiducial markers are commonly used in Image Guided Radiotherapy as landmarks to ensure patient setup accuracy and reproducibility. Good fiducials need to be radiopaque, yet not so much to create significant CT artifacts and disturb the tumor dose downstream. The latter is even more important in Proton radiotherapy to fully achieve its preciseness. The purpose of the study is to investigate several potential fiducial markers on their suitability for prostate proton therapy. METHODS In total, eight commercially available fiducial markers were compared in this study. These fiducials include: 3 types of VisiCoil (VC1-0.5×5 mm, VC2-0.35×5 mm, VC3-0.35×10 mm), BiomarC, FusionCoil, PolyMark, PointCoil, and FlexiCoil. We compared them in 4 ways. First, we scanned them using a GE RT16 scanner with a clinically used scanning protocol, and evaluated the degree of CT artifacts in two different water tank geometries; second, we evaluated visibility on kV images used for patient positioning, third, we compared their visibility on MRI. In addition, we looked at the perturbance of isodose lines distal to the fiducials in our treatment planning system. RESULTS Our preliminary study shows that PolyMark, BiomarC, VC2, and VC3 have the least CT artifacts of the group; FusionCoil, Flexicoil, PointCoil, and VC1 are most visible and PolyMark almost invisible on kV images; Evaluation of MRI visibility and dosimetric effects for these fiducials are in progress. CONCLUSION The VC1 and BiomarC seem to have the best performance in balance of CT artifacts and x-ray visibility. Further study on MRI visibility and dosimetric effect will be conducted to determine the overall performance of each fiducial and its suitability for clinical use in proton therapy.

SU‐E‐J‐76: Clinical Use of a Real‐Time Surface Image‐Guided Positioning and Tracking System in Proton Therapy

Topic of interest: Clinical applications of AlignRT 3-cameras real-time surface image-guided positioning system (IGPS) for positioning patients to reduce the number of X-ray images and tracking intra-fractional movements in proton therapy. PURPOSES To position patients and track the intra-fractional movements, the AlignRT system was implemented in proton incline-beam-line (IBL) at Procure Oklahoma-City center. METHODS The AlignRT3c system was configured near perpendicular to the gantry rotation for accommodating the X-ray IGPS. To evaluate positioning accuracy, more than 10 surfaces of each patient for ten patients with intracranial tumors were acquired after patients positioned by X-ray IGPS. Displacements between acquired surfaces and the reference surface taken at 1st day of treatment were examined. Intra-fractional movements with respiratory was studied with gated surface that allows setting the reference surface for patient at exhale during breathing. Intra-fractional movements due to respiratory were monitored on 10 sections of each patient for three patients with thoracic tumors. RESULTS Accuracy of positioning patient is 2.0 mm at both anterior-posterior and lateral directions, and is 3.5 mm in superior-inferior (SI) direction by aligning the surfaces of masks. Observed larger displacements along SI direction can be due to patients movements within the mask. Periodical displacements within 5 mm compared to its reference were seen for the three patients with thorax tumors. However, 10 mm sharp displacements with a few seconds were observed when patient moved the body. CONCLUSIONS We have implemented the first AlignRT3c IGPS for proton therapy for positioning patients within 2.0 mm, and successfully tracked intra-fractional respiratory motion during treatment after positioning patient.

SU‐E‐T‐45: Can We Model Proton Beam Output Factors Accurately without Measurements?

Purpose: Currently, no commercial protontreatment planning systems have been FDA‐approved for Monitor Unit (MU) prediction. Patient and field specific MU numbers and output factors are therefore determined by either measurements or in‐house modeling. The purpose of this project is to assess the modeling accuracy and determine in which cases measurements are necessary. Methods: We developed an in‐house analytical model to predict output factors for patient specific proton fields, taking into account the effect of proton range, modulation width, and field size on the dose output. A refined model, which incorporated more commissioning data and accounted for the effects of scanning area and field size, was developed later and retrospectively applied to the fields. Model‐predicted and measured output factors were compared for 1074 proton fields. Causes for the differenceswere analyzed for differences of 2.5% or more, which would be used for future improvement and determine in which cases where measurements are needed. Results: For the 1074 patient fields we analyzed, the refined model predicted output factors within 2% for 91.6% of fields, 2.5% for 96.9% for fields, and 3% for 99.7% of fields, compared to 85.2%, 93.6 and 98.2% respectively for the old model. Large differences between modeling and measurements (>2.5%) typically occurred for shallow range, large modulation, and/or irregular shaped proton fields. Causes for differences were analyzed, which mainly include dose rate dependence, off axis fields, data interpolation, and snout position effects. Conclusion: The study demonstrated that the analytical model can predict output factors accurately for most patient specific proton fields. For over 90% of proton fields, measurements may not be necessary, thus improving work flow and saving valuable proton beam time. It is possible to further improve modeling accuracy by incorporating more commissioning data, dose rate effects and other factors into the output factor model.