ESTRO 2021 Abstract Book

S131

ESTRO 2021

from scattered radiation and beam hardening influence the Hounsfield Unit accuracy. The aim of this study was to investigate a breast-specific stoichiometric CBCT calibration and to explore its effect on proton dose calculations in both phantom and patient scans. Materials and Methods A breast-specific stoichiometric calibration was performed to create a stopping power Hounsfield look up table (HLUT) using the Gammex Advanced Electron Density Phantom with the central part replaced by a low- density foam to simulate lungs. Fifteen different tissue substitute inserts were placed one-by-one in the phantom at the breast position (Fig.1) and CBCT images were acquired with the Varian ProBeam thorax imaging protocol (125 kVp, 2000 mAs, half fan bowtie-filter). The inserts ranged in electron density from lung to bone. Validation was performed with a standard breast plan (50 Gy(RBE)/25 fractions) on the CIRS thorax phantom with a breast of tissue equivalent epoxy material and on treatment planning data for one previously treated breast cancer patient. The planning CT (pCT) proton plan was recalculated on four repeat CTs (rCT), using the clinical CT HLUT, and CBCTs from the same day (thorax protocol, 1000 mAs), using the CBCT HLUT. Structures from the rCTs were transferred to rCBCTs through deformable registration and adjusted by a clinician. To take fixation differences between rCTs and rCBCTs into account, recalculations on the rCTs were performed with extra/missing tissue from the CBCTs overwritten with breast tissue or air, accordingly (rCT override).

Results The CBCT HLUT was close to the CT HLUT in the soft tissue region but deviated slightly for lung and bony tissues (Fig.1). For the phantom case, differences in DVH parameters between CT and CBCT ranged between - 0.7 and 1.2 Gy(RBE) with a median of -0.6 Gy(RBE) (Fig.2a), where negative values correspond to a lower calculated dose in the CBCTs. The largest difference was seen in mean dose for ipsilateral lung. For the patient case, the target coverage volume differences before override between rCTs and rCBCTs varied between -1.8% and 4.6% where the largest differences were observed in the coverage of the internal mammary nodes (IMN) (Fig.2b). These differences were reduced substantially for the rCT override which was also the case for the OARs. Conclusion Proton dose calculation on CBCT images, using a breast specific CBCT calibration, is promising for daily dose verification. This method is ready for clinical implementation using the same equipment and analysis as for CT calibration.