Feasibility of prostate offline adaptive radiation therapy using iterative reconstructed CBCT to optimize the ART workflow.

BACKGROUND: Adaptive Radiotherapy (ART) updates treatment plans based on newly acquired anatomical information to maintain dosimetric accuracy. Image-Guided Radiotherapy (IGRT) is essential for detecting anatomical changes and guiding replanning decisions. PURPOSE: This study evaluates the feasibility of using iterative Cone Beam CT (iCBCT) acquired on a TrueBeam system as a reference image dataset for dose calculation and offline ART in prostate radiotherapy. The approach avoids additional simulation CTs and reduces uncertainties from deformable image registration (DIR). MATERIALS AND METHODS: A CIRS electron-density phantom was embedded in solid-water slabs and surrounded by bolus to simulate varying patient-scanned volumes. HU-density calibration curves were generated for a standard prostate patient volume (sHU-density) and volume-specific conditions (vHU-density) and imported into Eclipse planning system along with the planning (pCT) calibration curve (pHU-density). Twenty prostate VMAT plans were retrospectively recalculated on registered iCBCT images using these curves. Dosimetric parameters (Dmean, D98%, D2%) and 3D gamma analysis (1%/1 mm, 10% threshold) assessed dose differences. An Alderson Rando phantom study was used to compared pCT-based (pHU-density) and iCBCT-based (sHU-density) calculations. RESULTS: HU-density curves were stable across scanned volumes, particularly for soft tissues. Dosimetric differences between sHU-density and vHU-density curves were < 0.5%, with gamma pass rates > 99%. Using the pHU-density curve on iCBCT introduced 1% systematic dose deviations. Phantom measurements confirmed strong agreement between pCT- and iCBCT-based calculations using an iCBCT-specific calibration curve. CONCLUSION: iCBCT with a dedicated HU-density calibration curve provides dose calculation accuracy comparable to planning CT, supporting reliable offline ART in prostate radiotherapy.