Purpose Dose computation using cone beam computed tomography (CBCT) images is inaccurate for the purpose of adaptive treatment planning. The main goal of this study is to assess the dosimetric accuracy of synthetic computed tomography (CT)‐based calculation for adaptive planning in the upper abdominal region. We hypothesized that deep learning‐based synthetically generated CT images will produce comparable results to a deformed CT (CTdef) in terms of dose calculation, while displaying a more accurate representation of the daily anatomy and therefore superior dosimetric accuracy. Methods We have implemented a cycle‐consistent generative adversarial networks (CycleGANs) architecture to synthesize CT images from the daily acquired CBCT image with minimal error. CBCT and CT images from 17 liver stereotactic body radiation therapy (SBRT) patients were used to train, test, and validate the algorithm. Results The synthetically generated images showed increased signal‐to‐noise ratio, contrast resolution, and reduced root mean square error, mean absolute error, noise, and artifact severity. Superior edge matching, sharpness, and preservation of anatomical structures from the CBCT images were observed for the synthetic images when compared to the CTdef registration method. Three verification plans (CBCT, CTdef, and synthetic) were created from the original treatment plan and dose volume histogram (DVH) statistics were calculated. The synthetic‐based calculation shows comparatively similar results to the CTdef‐based calculation with a maximum mean deviation of 1.5%. Conclusions Our findings show that CycleGANs can produce reliable synthetic images for the adaptive delivery framework. Dose calculations can be performed on synthetic images with minimal error. Additionally, enhanced image quality should translate into better daily alignment, increasing treatment delivery accuracy.

Purpose Recently a novel radiochromic sheet dosimeter, termed as PRESAGE sheets, consisting of leuco crystal violet dye and radical initiator had been developed and characterized. This study examines the dosimeter’s temporal stability and storage temperature dependence postirradiation, and its applicability for dose verification in three dimensions (3D) as a stack dosimeter. Methods PRESAGE sheets were irradiated using 6 MV photons at a dose range of 0–20 Gy with the change in optical density measured using a flatbed scanner. Following their irradiation, PRESAGE sheets were stored in different temperature environments (–18 °C, 4 °C, and 22 °C) and scanned at different time points, ranging from 1 to 168 h postirradiation, to track changes in measured signal and linearity of dose response. Multiple PRESAGE sheets were bound together to create a 12 × 13 × 8.7 cm3 film stack, with EBT3 film inserted between the sheets in the central region of the stack, that was treated using a clinical VMAT plan. Based on the results from the time and storage temperature study, two‐dimensional (2D) relative dose distribution measurements in PRESAGE were acquired promptly following irradiation at selected planes in the coronal, sagittal, and axial orientation of the film stack and compared to the treatment planning system calculations in their respective axes. Dose distribution measurements on the coronal axis of the stack dosimeter were also independently verified using EBT3 film. Results The dose response was observed to be linear (R2> 0.995) with sheets stored in colder temperatures retaining their signal and dose response sensitivity for extended periods postirradiation. Sheets stored in 22 °C environment should be measured within an hour postirradiation. Sheets stored in a 4 °C and −18 °C environment can be scanned up to 20‐ and 72 h postirradiation, respectively, while preserving the integrity of their dose response sensitivity and linearity of dose response within a mean absolute percent error of 2.0%. For instance, at 20 h postirradiation the dose response sensitivity for sheets stored in a −18 °C, 4 °C, and 22 °C temperature environment was measured to be 97%, 91%, and 77% of their original values measured within an hour postirradiation, respectively. The 2D gamma pass rate for central slices exceed 95% for PRESAGE film stack compared with treatment planning system on selected planes in the axial, coronal, and sagittal orientation and EBT3 film in the coronal orientation using a 2D gamma index of 2%/2mm. The gamma pass rate in comparing the calculated dose distribution with the measured dose distribution from PRESAGE‐LCV was observed to decrease in sheets scanned at later elapsed times postirradiation. In one example, the gamma pass rate for 2%/2mm criteria in the coronal plane was observed to decrease from 97.7% pass rate when scanned within an hour postirradiation to 92.1% pass rate when scanned at 20 h postirradiation under room temperature conditions. Conclusions This is the first study to demonstrate that the temporal s...

Materials/Methods: Thirty-three and 30 pairs of pre-treatment CT and first-fraction CBCT were selected from patients treated with proton therapy in our institution for Lung and H&N cancer, respectively. Deformable registration was performed to eliminate any potential setup variation between CT simulation and first fraction. The paired CT-CBCT datasets were then divided into training, validation and testing groups (lung patients: 24, 3 and 6, H&N patients: 22, 3 and 5 respectively). Two types of GAN models, including one generator with Orginal-Unet (Org-GAN) or Residual-Unet (Res-GAN) architecture, and one discriminator using a multi-layer convolutional neural net (CNN), were trained from scratch to predict CT from CBCT, with random deformation as data augmentation. Two training schemes were also tested, with two groups of patients trained separately or mixed together. All evaluations were performed on the testing group. The model-predicted synthetic CTs (synCT) were compared to the paired CTs in terms of Mean Absolute Error (MAE). The clinical treatment plans were recalculated on both synCT and paired CT using a commercial Monte-Carlo algorithm. Clinically relevant DVH parameters and g passing rates were utilized to quantify the dosimetric discrepancy. Results: The best MAE was achieved for Res-GAN model with the separately training scheme, 49.2AE6.7 and 49.6AE2.1 HU for Lung and H&N group, respectively. Compared to the reference doses calculated on the paired CT, the 3%/3mm and 2%/2mm g passing rates of the doses on Res-GAN synCT were 99.5AE0.4% and 95.3AE1.3 % for lung patients and 99.6AE0.4% and 97.0AE1.2% for H&N patients. The clinic relevant DVH parameters were within 1.0% for lung patients, except for cord D_1cc (1.9AE1.1%), mean dose of esophagus (1.3AE0.7%) and Trachea (1.1AE0.3%). For H&N patients, all DVH parameters agreed well within 0.5%, except for the mean dose of oral-cavity (2.7AE3.0%) and esophagus (2.3AE3.0%). Conclusion: With a relatively small number of training patients, the proposed Res-GAN model can predict synCT via CBCT with clinically acceptable dosimetric accuracy for proton treatment. The proposed method could be utilized to monitor potential proton dose deviation and trigger plan adaptation for lung and H&N patients.

Purpose/Objective(s): Treatment planning using a single isocenter for multiple brain metastases with conformal dose distribution is challenging in terms of plan quality, planner variance, and planning efficiency. Automated planning techniques are currently available to provide efficient planning and consistent plan quality. This study evaluates two automated treatment planning techniques for multiple brain metastases using a single isocenter. Materials/Methods: Eighteen patients with a total of 96 lesions who underwent treatment for multiple brain metastases (3-10 lesions) were included in this retrospective study. The clinical plans were originally generated with VMAT technique in a treatment planning system. Using the same beam geometry, these patients were replanned with a knowledgebased planning (KBP) SRS model based on 100 prior monoisocentric SRS plans. The same cohort of patients were replanned with the Multiple Brain Mets (MBM) SRS module in a commercially available SRS planning software using noncoplanar dynamic conformal arcs with the same numbers of couch kicks. Both techniques were designed to automatically generate volumetrically optimized plans for all metastases using a single isocenter. All the plans were optimized and calculated only once without further manual tuning. The MBM plans were imported into the TPS to compare with the corresponding KBP plans on the same platform. Plan evaluation metrics including the PTV coverage, conformity index (CI), gradient index (GI), total monitor units (MUs), planning time, brain V 12Gy , V 8Gy , and V 5Gy were recorded. Planning times were recorded from initial PTV assignment for optimization to the end of dose calculation. Comparisons of the KBP and MBM plans were performed using the two-tailed paired student t-test. Results: For KBP vs. MBM plans, PTV coverage mean AE standard deviation values were (99.2% AE 1.2%) vs. (98.9% AE 1.6%); CI were 1.5 AE 0.4 vs. 1.5 AE 0.2; GI were 3.0 AE 1.6 vs. 3.1 AE 1.9. Planning time mean AE standard deviation values were 19.2 AE 3.8 vs. 6.3 AE 1.2 minutes (p < 0.05); MUs were 2366 AE 1496 vs. 2892 AE 1828 for KBP and MBM plans, respectively (p < 0.05). Brain V 12Gy Z (64.9 AE 34.2) cc vs. (66.1 AE 50.7) cc; V 8Gy Z (172.6 AE 97.1) cc vs. (146.5 AE 109.9) cc; V 5Gy Z (422.8 AE 216.9) cc vs. (324.8 AE 205.5 cc) (p < 0.05). Conclusion: Both KBP and MBM automated planning techniques produced clinically acceptable plans and yielded comparable PTV coverage, CI, and GI values. In cases with irregularly shaped lesions, KBP plans provided better PTV coverage vs. MBM plans. KBP plans took significantly longer to plan but shorter treatment times due to fewer MUs required vs. MBM plans. Brain V 12Gy were comparable, but MBM plans spared healthy brain better vs. KBP plans in terms of brain V 8Gy and V 5Gy .

No abstract