B Guo

SU‐GG‐T‐38: Dosimetry and Inverse Treatment Planning for 3D Intensity Modulated Brachytherapy

Purpose: To propose a dosimetry algorithm for three‐dimensional (3D) intensity modulated brachytherapy (IMBT) and to develop an inverse treatment planning method applying the dosimetry algorithm. Method and Materials: A Matlab based prototype 3D IMBT treatment planning system was developed. The system consisted of three main components: (1) a comprehensive source commissioning method for intensity modulated sources based on Monte Carlo (EGSnrc) simulations; (2) a “modified TG43” (mTG43) dose calculation algorithm for IMBT dosimetry; (3) an inverse IMBT treatment planning method based on Dose Volume Histogram (DVH) or DoseSurface Histogram (DSH) constraints and simulated annealingoptimization algorithm. The system was applied for planning of an intracavitary accelerated partial breast irradiation (APBI) case treated with Xoft Axxent electronic brachytherapy. Plan quality, planning and delivery time of the IMBT plan were compared with the original plan used for the patients treatment. Results: For the patient studied, IMBT plan showed better plan quality compared with the original plan. With similar coverage to the target, high dose region V200 was decrease by 16.1%. Maximum doses to skin and ribs were reduced by 56 cGy and 104 cGy in one fraction respectively. Mean dose to ipsilateral and contralateral breasts and lungs were also slightly reduced by IMBT. Conclusion: Application of three‐dimensional intensity modulation in brachytherapytreatment planning is both feasible and promising. 3D IMBT improves the quality of APBI brachytherapy treatment plan, increasing dose uniformity in target and reducing the dose to critical structures.

SU‐FF‐T‐05: A Monte Carlo Based Dose Calculation and Evaluation Toolkit for Electronic Brachytherapy: Feasibility of IMBT

Purpose: To develop a Monte Carlo based dose calculation and evaluation toolkit for electronic brachytherapy sources, capable of simulating Intensity Modulated Brachytherapy (IMBT). Material and methods: This toolkit used Monte Carlo code EGS4 to calculate the dose distribution for a treatment in a realistic virtual human phantom converted from the patients CTimages and contours. An in‐house Matlab program was developed to analyze the dose distribution and generate DVHs and isodoses. Results: The system was benchmarked by comparing the calculated radial dose functions (in water) with experimental data published. Difference was within 3%. A typical intracavitary accelerated partial breast irradiation (APBI) treatment plan using Xoft Axxent electronic brachytherapy source was simulated using this toolkit. DVHs and isodoses revealed that the dose to the ribs were high in this plan due to proximity of balloon to the chest wall and the high absorption coefficient of bone to low energy X‐rays. A simple IMBT plan using partial block could conform the isodose distribution to the target, reduce the dose to ribs and chest wall without compromising the dose homogeneity to the target or increasing the dose to other critical structures. Conclusion: A Monte Carlo based dose calculation and evaluation system was developed for electronic brachytherapy sources. Benchmarked through published data, this system is capable of producing reliable and detailed dose distributions for both isotropic and intensity modulated sources. The feasibility of intensity modulation in improving the plan quality was proved.

SU‐FF‐T‐411: Boundary Study of Bragg Peak Shift and Bragg Peak Degradation in Proton Dose Calculation

Purpose: The purpose of this work is to study the Bragg peak shifts and degradation caused by density and boundary changes in proton beam dose calculation Method and Material:Proton beam delivery provides promising dose characteristics as radiation dose can conform tightly to tumor while sparing surrounding healthy tissues. Proton particles deposit energy in a narrow range around the Bragg peak and as such the dose calculation is more challenging for that the Bragg peak is sensitive to tissue density, tissue composition and organ boundaries along the proton track path. We simulated a few scenarios to study the proton Bragg peak shift due to density and Bragg peak degradation due to change and boundary changes. The calculation of the three dimension dose matrix was performed using a 2 × 2 × 1 mm3 voxels in the depth peak dose range in water phantom after some rough simulation for the dose peak estimation. Results: Bragg peak shift at the iso‐center slice were found to follow a linear relationship with the density of heterogeneity insert based on our simulations with density ranging [0.4 2.0] g/cm∧3 which we studied. Bragg peak degradation and proton dose changed significantly due to low density and small beams size. Proton dose degraded when high energy proton beam irradiated to low density material.Proton dose degraded also when small beam with beam radius at several mm range. Conclusion:Proton dose calculation depends on many factors as Bragg peak is sensitive to tissue density and composition. Besides that, there exist several scenarios causing Bragg peak shift due to density change, causing Bragg peak degradation due to low density and small proton beam. The Monte Carlo simulation is a very accurate solution to provide precise dose distribution in inhomogeneous structures by simulating transport and energy deposition.

TU‐G‐BRC‐02: Predictive Tracking for Real‐Time Deformable Motion Using a Dynamic Virtual Patient Model

Purpose: To develop a novel tracking technique for intensity modulated radiation therapy(IMRT) based on dynamic virtual patient modeling. Method and Materials: Predictive tracking uses a dynamic virtual patient model to account for the three‐dimensional deformable motion of the target during realistic IMRT deliveries. The virtual patient model was developed based on the 4DCT images and the structure contours delineated on different respiratory phases. With a real‐time breathing curve, the virtual patient model could be updated to a real‐time model which would then be used to predict the motion of the target, to modify the IMRT leaf sequence on‐line to account for the target motion, and to calculate the delivery dose of the IMRT plan with consideration of the interplay effect. A lung case was chosen to demonstrate predictive tracking. The quality of the predictive tracking plan was compared with two commonly used 4D IMRTtreatment planning techniques: maximum intensity projection (MIP) and planning on individual phases (IP). Results: For a realistic breathing motion, predictive tracking offered better plan quality than MIP and IP. PTV V97 was 90.4% for a MIP plan, 88.6% for an IP plan and 94.1% for a predictive tracking plan. Lung V20 was 20.1% for the MIP plan, 17.8% for the IP plan and 17.5% for the predictive tracking plan. Conclusion: Predictive tracking based on dynamic virtual patient modeling is a novel 4D planning/delivery technique which is capable of managing the real time 3D deformable motion of target without increasing the workload of treatment planning or time of delivery. For a realistic breathing motion with consideration of the interplay effect in beam delivery, predictive tracking gave better plan quality than IP and MIP. Conflict of interest: Research supported by grant R01 LM009362.

SU‐GG‐T‐250: 3D Dose Reconstruction for Delivery Quality Assurance (DQA) from Multiple 2D Planes Using the OCTAVIOUS Phantom

Purpose: To perform a 3D dose reconstruction for delivery quality assurance (DQA) from multiple 2D planes using the OCTAVIOUS phantom. Method and Materials: Ten RapidArc patient treatment plans of different sites were delivered on two phantoms. Two different DQA plans were delivered for each of the 10 patients: 1) OCTAVIOUS phantom and 2) 30×30×30 cube solid water phantom in which the detector array (Seven29) was placed. The corresponding 3D dose of each of the DQA plans was exported. All DQA plans were delivered by means of a NovalisTX with the HD120 MLC. For the solid water phantom, the same plan was delivered six times with the array varying in the coronal plane in increments of 0.5cm. For the OCTAVIOUS, the same plan was delivered four times by rotating the phantom in 45° increments along its longitudinal axis. An in‐house MATLAB code was used to read the planar dose information. A linear reconstruction was performed for the cube phantom while a circular reconstruction was used for the OCTAVIOUS. Dose statistics per plane were obtained to validate the reconstruction method. Results: Both interpolation methods showed good agreement to the planned dose distribution in the high dose region (<1.0%) but showed discrepancies in the low dose region. A DAH comparison shows good agreement for the sagittal and coronal planes but demonstrates some discrepancies in the transversal plane. Conclusion: A simple linear interpolation method is able to predict good matching in the high dose region between the reconstructed dose and the planned dose. This technique is a good starting point to establish a benchmark in the level of manipulation necessary to obtain good 3D dose delivery quality assurance using current technology. Conflict of Interest: Research Sponsored by PTW Company

SU-GG-J-96: Statistical Analysis of the Correlation between Breathing Characteristics with Patient Parameters

Purpose: Breathing curve of lung patient is an important feature for tumor tracking and 4D delivery. The correlation of external breathing curve characteristics with internal tumor location and with the biological features of the patient is of large interest. In this study, we are aiming to propose a method for analyzing the parameters of breathing curves and study the correlation of breathing characteristics with features of patients. Method and Materials: Piecewise cosine functions were used to fit breathing curves and determine the amplitude, period and baseline of each breathing cycle. Mean value and standard deviation of amplitude, period and baseline over the breathing curves were determined as the “characteristics” of a breathing curve. Statistical analysis was performed to correlate these characteristics with patient parameters including age, gender and tumor site base on a total of 305 breathing curves from 158 patients acquired from Real‐time Position Management (RPM) system. Pearson correlation method was used to determine the correlation of age with breathing parameters. Two‐sided unpaired t‐test was used to study whether gender or presence of tumor in lung affect the breathing characteristics significantly. Results: Age has no correlation with breathing characteristics studied. Gender affects both the mean (p = 0.971) and standard deviation (p = 0.425) of breathing period. The effect of gender on breathing amplitude or baseline variation was insignificant. Presence of lungtumor affects the mean amplitude (p = 0.317) and period (p = 0.384). The baseline variation was also affected slightly (p = 0.860). But the irregularity of amplitude and period was irrelevant to the tumor locations. Conclusion:Statistical analysis of large number of breathing curves revealed that parameters of patient may affect the breathing curve statistics thus affecting the management of tumor motion in radiation therapy. Conflict of interest: Research supported by grant NIH/NLM R01LM009362

SU‐FF‐J‐107: 4D Predictive Patient‐Specific Anatomical Model Based On 4D CT Data: A Feasibility Study

Purpose: To construct a patient‐specific 4D predicting geometry model based on 4D CT data. Material and methods: Patients contours were re‐sampled and reconstructed using Non‐Uniform Rational B‐Spline (NURBS) surface. The breathing curve associated with the patient was predicted using Kalman filter and fitted using precise cosine functions. The 4D NURBS model was generated using 3D NURBS surface models in different phases. Results: The percent differences of the volume and the distances of the COM (center of mass) for all structures in phase P0 and P50 shows the model accurate within 5% for volume and 1 mm in three x, y, z directions. With the expense of sampling and reconstructing time, the percentage difference can be further improved. Conclusion: This study demonstrated the feasibility of constructing a patient‐specific 4D predicting geometry model based on 4D CT data. The model was controlled by the patients breathing curve. Since the breathing curve can be predicted using the Kalman filter and approximated using the cosine function, the model is predictive of patients lung motion. NURBS control points can reconstruct the anatomical deformations precisely and quickly with affordable computing power and time, therefore, the model has the potential for real‐time controlling and guiding of the radiation delivery.

MO‐E‐AUD B‐07: SAF Values for Internal Electron Emitters Calculated for the RPI‐P Pregnant‐Female Models Using Monte Carlo Methods

Purpose: to calculate specific absorbed fraction (SAF) values for internal electron emitters based on more realistic RPI‐P serial pregnant female models. Method and Materials: The RPI‐P series pregnant‐female models developed by Xu and coworkers were used for Monte Carlo simulation. Those models are based on boundary‐representation method for organ delineation. The image sources are from clinical CTimage, VIP‐Man image, and public domain images. The pregnant woman models, RPI‐P3, RPI‐P6, and RPI‐P9, were implemented into a previously developed Monte Carlo user code, EGS4‐VLSI. In this study, internal electron emitters were considered for the following energies: 10, 15, 20, 30, 50, 100, 200, 500, 1000, 1500, 2000, and 4000 keV. SAF values to the fetus were calculated for each of these energies involving 35 source organs.Results and Discussion: SAF factors from source organs to the fetus have been calculated for all the three pregnant female models. Results show that electron SAF values follow linear relationship as equation: log (SAF(fetus ← source) = A ⋅ log (E) + B , where E is the electron energy, A and B are coefficients. A and B coefficients were calculated. R2 coefficient, the determination for the linear relationship, is ranging from 0.90∼1.00 except source organ=heart for RPI‐P3 model. It means the linear relationship between log(SAF) and log(E) is fitting well. Conclusion: SAF values have been derived based on a new developed RPI‐P series pregnant‐female models using Monte Carlo method. For electron emitters ranging from 10 keV to 4000 keV, the log(SAF) and log(energy) relationship can be approximated by linear function.

SU‐GG‐T‐41: Dosimetric Impacts On Tissue Homogeneity Corrections in Electronic Brachytherapy

Purpose: With increasing interests in utilizing electronic brachytherapy in accelerated partial breast irradiation, the lack of tissue inhomogeneity correction in the dose calculation draws clinical attentions. In this study, a Monte Carlo based dose evaluation was assessed in a human anatomy phantom to demonstrate these dosimetricimpacts.Method and Materials: The dosimetriccharacteristics of Xoft S700 Axxent™ X‐ray source (Xoft, Fremont,CA) were evaluated using EGS4 Monte Carlo code to verify the calculation method and source parameter accuracy. A virtual female human phantom utilizing all clinical parameters was used in the simulation. A balloon (diameter = 5 cm) was inserted in the right breast of the phantom. A 1 cm expanded volume from the balloon surface was designed to define as planning target volume (PTV). The prescribed dose was 3.4 Gy to 5 selected points (anterior, posterior, right lateral, superior and inferior) of the PTV. Dose,dose volume histograms of the PTV and normal tissues, isodose distributions in axial, sagittal and coronal planes crossing the center of the balloon were evaluated. Results: Using measured 50 kVp source spectrum provided by Xoft Inc, the simulated radial dose function agrees well with the published data within 3%. The simulation shows clear dose uniformity with 90.6% of the prescription dose enclosing PTV. The two dimensional sagittal dose distributions show high skin and rib doses with reasonably acceptable doses to all other surrounding normal tissues. When accounted for the tissue heterogeneity, changes in the isodose distribution were observed. Conclusion: This has been our first EGS4 Monte Carlo simulation study for the Xoft electronic brachytherapy system. Preliminary results demonstrate the dosimetricimpact especially on the surrounding normal tissues, when tissue inhomogeneity consideration is taken into account.

SU‐GG‐T‐346: Accurate Simulation for Therapeutic Protons Range‐Energy and Range‐Density Tables

Purpose: The purpose of this work is to generate accurate range‐energy and range‐density tables for therapeuticprotons using Monte Carlo simulation in water phantom. Method and Materials The percentage depth dose and the depth peak dose was calculated using Monte Carlo simulation for protons. The simulation was based on Monte Carlo code, MCNPX in water phantom with protons energy ranging from 40 MeV increasing by 10 to 250 MeV. The calculation of the three dimension dose matrix was performed using a 2 × 2 × 1 mm∧3 voxels in the depth peak range in water phantom. The pencil beam was used and the source was uniformly distributed in area with beam diameter as 20 mm. A 3,000,000 history was used in each individual simulation with uncertainty controlled less than 1.0%. Results: The total 22 data points of simulated depth peak were compared to two widely used proton range‐energy tables from International Commission on Radiological Units and Measurements (ICRU) Report 49 and from Data Nuclear Data Tables (DNDT). The simulated range‐energy curve matches very well with ICRU and DNDT with coefficient of determination equal to 1. The difference of the simulation and measurements is 1.17% at maximum and 0.70% at average compared to ICRU, 1.15% and 0.37% compared to DNDT respectively. The relation of range‐density was determined with water density ranging from 0.4 to 2.0 g/cm^3.The exponential relation of range‐density gives very closely matched estimation to the simulated value with the coefficient of determination as 0.9999. Conclusion: The simulated proton range‐energy matches very well to two widely used measurements. The accurate Monte Carlo simulation provides a close benchmark for existed measurements. The simulation offers the chance to replace the infinite experiments for each detailed energy and density. The simulation also expands the feasibility of accurate simulation in heterogeneity.