Q. Xu, H. Han, and L. Xing; Department of Radiation Oncology, Stanford. University, Stanford, CA. Purpose/Objective(s): Compressed sensing (CS)-based ...
E688
International Journal of Radiation Oncology Biology Physics
3685
necessity of increasing the immobilization and localization during scanning, planning, and treatment delivery. Materials/Methods: A retrospective analysis of 25 patients with 5 treatment plans each generated in a treatment planning system looked into the applicator’s placement in regard to the organs at risk. Motion possibilities for each applicator intra- and interfractionation with their dosimetric implications were covered and measured in regard with their dose variance. The localization immobilization devices used were assessed for the capability to prevent motion before and during the treatment delivery. Results: We focused on the 100% isodose on central axis and a 150 displacement due to possible rotation analyzing the dose variations to the bladder and rectum walls. The average dose variation for bladder was 15% of the accepted tolerance, with a minimum variance of 11.1% and a maximum one of 23.14% on the central axis. For the off axis measurements we found an average variation of 16.84% of the accepted tolerance, with a minimum variance of 11.47% and a maximum one of 27.69%. For the rectum we focused on the rectum wall closest to the 120% isodose line. The average dose variation was 19.4%, minimum 11.3% and a maximum of 34.02% from the accepted tolerance values. Conclusion: New and improved immobilization devices are recommended. For interfractionation, localization devices are recommended in place with consistent planning in regard to the initial fraction. Many of the present immobilization devices produced for external radiation therapy can be used to improve the localization of HDR applicators during transportation of the patient and during treatment. Author Disclosure: M. Shojaei: None. N. Dumitru: None. S. Chandrasekara: None. J. Pinder: None. M. Hyvarinen: None. C. Curley: None. S. Pella: None.
Initial Experience With Directional Low-Dose-Rate Intraoperative Brachytherapy S.X. Cavanaugh,1 D.J. Rothley,2 J.S. Dick,2 J.W. Swanson,3 and K. Watkins1; 1Cancer Treatment Centers of America, Newnan, GA, 2 Landauer Medical Physics, Newnan, GA, 3Landauer Medical Physics, Glenwood, IL Purpose/Objective(s): This study investigated the feasibility of using intraoperative directional brachytherapy for treatment of squamous cell carcinoma of the anus. The patient had received 2 prior courses of external beam therapy of 55.8 Gy in 2014 and 30 Gy in 2015. Due to the increased risk of additional external radiation, brachytherapy was considered as a treatment option. Materials/Methods: A commercially available flexible, bioabsorbable polymer membrane embedded with an array of discrete 103Pd sources was used for the implant. The 103Pd sources were spaced 8 mm apart on a 5x15 cm sheet yielding 108 sources. Unidirectional dose was achieved by a 0.05-mm thick gold disk-shaped foil on the reverse side of each source. A dose of 120 Gy at 5-mm depth was prescribed, dictating an air kerma of 3.0 U for each source. 120 Gy was selected because that dose has been used successfully to treat adenocarcinoma of the prostate with LDR brachytherapy. After the abdominoperineal resection, the entire sheet was placed on the treatment area in the patient to determine the needed dimensions. The polymer sheet was then removed and easily cut to size with scissors leaving 51103Pd sources remaining. The sheet was manufactured with holes in the membrane located between each source, through which a surgical tack could have been inserted to attach the sheet. However the surgeon elected to use sutures for attachment in a concave shape to the pelvic sidewall. The sheet was cut to cover the high-risk surgical bed and specifically a 3x4 cm area of muscle invasion and involved margins. The sheet was sutured with the radioactive side in contact with the pelvic sidewall and the gold, shielded side facing away. The surgical team completed the procedure and the patient recovered without any problems. Results: One week after the implant the patient received a CT scan in the radiation oncology department. The images were transferred to the brachytherapy treatment planning system for a post plan. The treatment plan indicated that the sources remained in position in a concave array pattern. Due to the dose fall-off of 103Pd, the calculated dose to critical structures was minimized. Conclusion: The surgical implant of the sheet proceeded as expected with no complications. The post plan indicated that the sheet remained in position with the radioactive side contacting the treatment area. The dosevolume histogram demonstrated good coverage with minimal dose to adjacent critical structures. Directional LDR intraoperative brachytherapy is a feasible alternative for retreatment of the high-risk margin of resection in pelvic malignancies. Author Disclosure: S.X. Cavanaugh: None. D.J. Rothley: None. J.S. Dick: None. J.W. Swanson: None. K. Watkins: None.
3686 The Importance of Localization and Immobilization in Treatment for Endometrial Cancer With High-Dose-Rate Brachytherapy Using Multilumen Cylinders M. Shojaei,1 N. Dumitru,1 S. Chandrasekara,1 J. Pinder,1 M. Hyvarinen,2 C. Curley,1 and S. Pella3; 1Florida Atlantic University, Boca Raton, FL, 2 University of Toronto, Toronto, ON, Canada, 3South Florida Radiation Oncology LLC, Boca Raton, FL Purpose/Objective(s): High-dose-rate (HDR) brachytherapy is a highly localized mode of radiation therapy that has a very sharp dose falloff. Thus one of the most important parts of the treatment is the immobilization. The smallest movement of the patient or applicator can result in dose variation to the surrounding tissues as well as to the tumor to be treated. Our purpose is to revise the MML cylinder treatments and their localization challenges. Since every millimeter of misplacement counts, the study will look into the
3687 Augmenting Compressed Sensing-Based CT/CBCT Image Reconstruction with Introduction of a Region-Specific Regularization Framework Q. Xu, H. Han, and L. Xing; Department of Radiation Oncology, Stanford University, Stanford, CA Purpose/Objective(s): Compressed sensing (CS)-based reconstruction, in which an objective function consisting of the data fidelity and penalty terms is optimized iteratively, has gained popularity in the past decade. It is particularly useful in radiation oncologyerelated applications because of its superior ability in image reconstruction under the condition of sparse data and low x-ray dose. However, most methods use a single regularization parameter to balance the fidelity and penalty in all regions of the object. In reality, noise and structural details vary from region to region, and a region-specific regularization would enable us to have much improved control of the image quality across the object. Development of such a CS framework with regionally optimized regularization is the purpose of this work. Materials/Methods: We adaptively select the regularization parameter according to the region-specific noise. In our workflow, we first configured one reconstruction without regularization to estimate the noise distribution by calculating the variance of neighboring pixels. Then a fuzzy clustering method was applied on the obtained noise to proportionally assign the membership of each pixel into 3 clustering groups which correspond to regions with small, medium, and large noises, respectively. Following that, we performed 3 reconstructions with small, medium, and large regularization parameters, respectively, according to each of these 3 noise levels. Finally these 3 reconstruction results were fused together by using the weighting coefficients (i.e., the proportional fuzzy membership) estimated in the fuzzy clustering step. A human thorax dataset was used to evaluate the proposed method. We used structure similarity index (SSIM) and root mean squared error (RMSE) to quantitatively evaluate the results. Results: Under the conventional CS reconstruction with a single regularization parameter, we failed to obtain a satisfactory image quality simultaneously in all regions. More noise was observed in the posterior region than other regions due to the nonuniform of projections’ noise. The proposed method selected the regularization parameter according to the
