Scientific Computing Algorithms to Learn Enhanced Scalable Surrogates for Mesh Physics
Explore the source record for details and available documents.
Engineering topics
Publications and source records attributed to Xu, Z.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Purpose: Normal organ segmentation is one time-consuming and labor-intensive step for lung radiotherapy treatment planning. The aim of this study is to evaluate the performance of a multi-atlas based segmentation approach for automatic organs at risk (OAR) delineation. Methods: Fifteen Lung stereotactic body radiation therapy patients were randomly selected. Planning CT images and OAR contours of the heart - HT, aorta - AO, vena cava - VC, pulmonary trunk - PT, and esophagus – ES were exported and used as reference and atlas sets. For automatic organ delineation for a given target CT, 1) all atlas sets were deformably warped to the target CT, 2) the deformed sets were accumulated and normalized to produce organ probability density (OPD) maps, and 3) the OPD maps were converted to contours via image thresholding. Optimal threshold for each organ was empirically determined by comparing the auto-segmented contours against their respective reference contours. The delineated results were evaluated by measuring contour similarity metrics: DICE, mean distance (MD), and true detection rate (TD), where DICE=(intersection volume/sum of two volumes) and TD = {1.0 - (false positive + false negative)/2.0}. Diffeomorphic Demons algorithm was employed for CT-CT deformable image registrations. Results: Optimal thresholds were determined to be 0.53 for HT, 0.38 for AO, 0.28 for PT, 0.43 for VC, and 0.31 for ES. The mean similarity metrics (DICE[%], MD[mm], TD[%]) were (88, 3.2, 89) for HT, (79, 3.2, 82) for AO, (75, 2.7, 77) for PT, (68, 3.4, 73) for VC, and (51,2.7, 60) for ES. Conclusion: The investigated multi-atlas based approach produced reliable segmentations for the organs with large and relatively clear boundaries (HT and AO). However, the detection of small and narrow organs with diffused boundaries (ES) were challenging. Sophisticated atlas selection and multi-atlas fusion algorithms may further improve the quality of segmentations.
Purpose: To compare three-dimensional conformal radiotherapy (3D CRT) and volumetric-modulated arc therapy (VMAT) in lung stereotactic body radiation therapy (SBRT) Methods: A retrospective study of clinically treated lung SBRT cases treated between 2010 and 2015 at our hospital was performed. All treatment modalities were included in this evaluation (VMAT, 3D CRT, static IMRT, and dynamic conformal arc therapy). However, the majority of treatment modalities were either VMAT or 3D CRT. Treatment times of patients and dosimetric plan quality metrics were compared. Treatment times were calculated based on the time the therapist opened and closed the patient’s treatment plan. This treatment time closely approximates the utilization time of the treatment room. The dosimetric plan quality metrics evaluated include ICRU conformity index, the volume of 105% prescribed dose outside PTV, the ratio of volume of 50% prescribed dose to the volume of PTV, the percentage of maximum dose at 2 cm away from PTV to the prescribed dose, and the V20 (percentage of lung volume receiving 20 Gy or more). Results: Treatment time comparisons show that on average VMAT has shorter treatment times than 3D CRT. Dose conformity, defined by the ICRU conformity index, and high dose spillage, defined by the volume of 105% dose outside the PTV, is reduced when using VMAT compared to 3D CRT. V20 and intermediate dose spillage/fall-off metrics of VMAT and 3D are not significantly different. Conclusion: Clinically treated lung SBRT cases indicate VMAT is superior to 3D with regard to shorter treatment times, plan dose conformity, and plan high dose spillage.
Purpose: Commercial dose calculation algorithms incorporate a single DLG value for a given beam energy that is applied across an entire treatment field. However, the physical processes associated with beam generation and dose delivery suggest that the DLG is not constant. The aim of this study is to evaluate the variation of DLG among all leaf pairs, to quantify how this variation impacts delivered dose, and to establish a novel method to correct dose distributions calculated using the approximation of constant DLG. Methods: A 2D diode array was used to measure the DLG for all 60 leaf pairs at several points along each leaf pair travel direction. This approach was validated by comparison to DLG values measured at select points using a 0.6 cc ion chamber with the standard formalism. In-house software was developed to enable incorporation of position dependent DLG values into dose distribution optimization and calculation. The accuracy of beam delivery of both the corrected and uncorrected treatment plans was studied through gamma pass rate evaluation. A comparison of DVH statistics in corrected and uncorrected treatment plans was made. Results: The outer 20 MLC leaf pairs (1.0 cm width) have DLG values that are 0.32 mm (mean) to 0.65 mm (maximum) lower than the central leaf-pair. VMAT plans using a large number of 1 cm wide leaves were more accurately delivered (gamma pass rate increased by 5%) and dose coverage was higher (D100 increased by 3%) when the 2D DLG was modeled. Conclusion: Using a constant DLG value for a given beam energy will result in dose optimization, dose calculation and treatment delivery inaccuracies that become significant for treatment plans with high modulation complexity scores delivered with 1 cm wide leaves.
Purpose: The commercially available Leipzig-style Cone for High Dose Rate (HDR) Brachytherapy has a steep depth dose curve and a non-uniform dose distribution. This work shows the performance of a Ring Surface Applicator created using a 3D printer that can generate a better dose distribution. Calculated doses were verified with film measurement. Methods: The water equivalent red-ABS plastic was used to print the Ring Surface Applicator which hosts three catheters: a center piece with a straight catheter and two concentric rings with diameters of 3.5 and 5.5 cm. Gafchromic EBT2 film, Epson Expression 10000 flatbed scanner, and the online software at radiochromic.com were used to analyze the measured data. 10cm×10cm piece of film was sandwiched between two 15×10×5cm3 polystyrene phantoms. The applicator was positioned directly on top of the phantom. Measurement was done using dwell time and positions calculated by Eclipse BrachyVision treatment planning system (RTP). Results: Depth dose curve was generated from the plan and measurement. The results show that the measured and calculated depth dose were in agreement (<3%) from surface to 4mm depth. A discrepancy of 6% was observed at 5 mm depth, where the dose is typically prescribed to. For depths deeper than 5 mm, the measured doses were lower than those calculated by Eclipse BrachyVision. This can be attributed to a combination of simple calculation algorithm using TG-43 and the lack of inhomogeneity correction. Dose profiles at 5 mm depth were also generated from TPS calculation and measured with film. The measured and calculated profiles are similar. Consistent with the depth dose curve, the measured dose is lower than the calculated. Conclusion: Our results showed that the Ring Surface Applicator, printed using 3D printer, can generate more uniform dose distribution within the target volume and can be safely used in the clinic.
Purpose: This study is to use plan parameters optimization (Dose rate, collimator angle, couch angle, initial starting phase) to improve the performance of conformal arc radiotherapy plans with motion tracking by increasing the plan performance score (PPS). Methods: Two types of 3D conformal arc plans were created based on QUASAR respiratory motion phantom with spherical and cylindrical targets. Sinusoidal model was applied to the MLC leaves to generate motion tracking plans. A MATLAB program was developed to calculate PPS of each plan (ranges from 0–1) and optimize plan parameters. We first selected the dose rate for motion tracking plans and then used simulated annealing algorithm to search for the combination of the other parameters that resulted in the plan of the maximal PPS. The optimized motion tracking plan was delivered by Varian Truebeam Linac. In-room cameras and stopwatch were used for starting phase selection and synchronization between phantom motion and plan delivery. Gaf-EBT2 dosimetry films were used to measure the dose delivered to the target in QUASAR phantom. Dose profiles and Truebeam trajectory log files were used for plan delivery performance evaluation. Results: For spherical target, the maximal PPS (PPSsph) of the optimized plan was 0.79: (Dose rate: 500MU/min, Collimator: 90°, Couch: +10°, starting phase: 0.83π). For cylindrical target, the maximal PPScyl was 0.75 (Dose rate: 300MU/min, Collimator: 87°, starting phase: 0.97π) with couch at 0°. Differences of dose profiles between motion tracking plans (with the maximal and the minimal PPS) and 3D conformal plans were as follows: PPSsph=0.79: %ΔFWHM: 8.9%, %Dmax: 3.1%; PPSsph=0.52: %ΔFWHM: 10.4%, %Dmax: 6.1%. PPScyl=0.75: %ΔFWHM: 4.7%, %Dmax: 3.6%; PPScyl=0.42: %ΔFWHM: 12.5%, %Dmax: 9.6%. Conclusion: By achieving high plan performance score through parameters optimization, we can improve target dose conformity of motion tracking plan by decreasing total MLC leaf travel distance and leaf speed.
Purpose: The purpose of this project is to demonstrate that a non-expensive 3D-printer can be used to manufacture a 3D-bolus for external beam therapy. The printed bolus then can be modeled in our treatment planning system to ensure accurate dose delivery to the patient. Methods: We developed a simple method to manufacture a patient-specific custom 3Dbolus. The bolus is designed using Eclipse Treatment Planning System, contoured onto the patients CT images. The bolus file is exported from Eclipse to 3D-printer software, and then printed using a 3D printer. Various tests were completed to determine the properties of the printing material. Percent depth dose curves in this material were measured with electron and photon beams for comparison to other materials. In order to test the validity of the 3D printed bolus for treatment planning, a custom bolus was printed and tested on the Rando phantom using film for a dose plane comparison. We compared the dose plane measured on the film to the same dose plane exported from our treatment planning system using Film QA software. The gamma-dose distribution tool was used in our film analysis. Results: We compared point measurements throughout the dose plane and were able to achieve greater than 95% passing rate at 3% dose difference and 3 mm distance to agreement, which is our departments acceptable gamma pixel parameters. Conclusion: The printed 3D bolus has proven to be accurately modeled in our treatment planning system, it is more conformal to the patient surface and more durable than other bolus currently used (wax, superflab etc.). It is also more convenient and less costly than comparable bolus from milling machine companies.
Purpose: The aim of this study is to evaluate the variation of dosimetric leaf-gap (DLG) along the travel path of each MLC leaf pair. This study evaluates whether the spatial variations in DLG could cause dose differences between TPS-calculated and measured dose. Methods: The 6MV DLG values were measured for all leaf pairs in the direction of leaf motion using a 2-D diode array and 0.6cc ion chamber. These measurements were performed on two Varian Linacs, employing the Millennium 120-leaf MLC and a 2-D-DLG variation map was created via in-house software. Several test plans were created with sweeping MLC fields using constant gaps from 2mm to 10mm and corrected for 2-D variation utilizing in-house software. Measurements were performed utilizing the MapCHECK at 5.0cm depth for plans with and without the 2-D DLG correction and compared to the TPS calculated dose via gamma analysis (3%/3mm). Results: The measured DLGs for the middle 40 MLC leaf pairs (0.5cm width) were very similar along the central superior-inferior axis, with maximum variation of 0.2mm. The outer 20 MLC leaf pairs (1.0cm width) have DLG values from 0.32mm (mean) to 0.65mm (maximum) lower than the central leaf-pair, depending on off-axis distance. Gamma pass rates for the 2mm, 4mm, and 6mm sweep plans increased by 23.2%, 28.7%, and 26.0% respectively using the 2-D-DLG correction. The most improved dose points occur in areas modulated by the 1.0cm leaf-pairs. The gamma pass rate for the 10mm sweep plan increased by only 7.7%, indicating that the 2D variation becomes less significant for dynamic plans with larger MLC gaps. Conclusion: Fluences residing significantly off-axis with narrow sweeping gaps may exhibit significant variations from planned dose due to large differences between the true DLG exhibited by the 1.0cm leaf-pairs versus the constant DLG value utilized by the TPS for dose calculation.
N/A
Not Available