Abstract
BACKGROUND: HyperSight CBCT offers rapid image acquisition and enhanced image quality, supporting its integration into online adaptive radiotherapy (OART). However, its performance under respiratory-induced tumor motion requires further evaluation. PURPOSE: This study evaluates the feasibility of using rapid-acquisition HyperSight CBCT for internal target volume (ITV) delineation and dose calculation in the setting of respiratory motion, with implications for OART workflows. METHODS: A dynamic thoracic phantom simulating superior-inferior tumor motion (5-25 mm amplitude from the motion center, 6-second cycle time) was imaged using Varian's HyperSight CBCT on the Ethos platform. Four reconstruction algorithms were evaluated: filtered back projection (FDK), iterative CBCT (iCBCT), iCBCT with Acuros, and metal artifact reduction (MAR). Experiments included variation of cycle times (4, 6, and 8 s), breathing patterns (Cos(6), sinusoidal, and hysteresis), and repeat acquisitions to assess the impact of scan initiation time relative to the respiratory cycle. In addition to fast scans, slow acquisitions (60-second cycle) were performed for further evaluation. Reference datasets included 4DCT maximum intensity projection (MIP) and average intensity projection (AIP) for ITV delineation and HU comparison. Volumetric modulated arc therapy plans were created on AIP images and recalculated on HyperSight images. Dose distributions were compared using gamma analysis (1%/1 mm, 10% threshold). RESULTS: For the Cos(6) breathing pattern with a 6-second cycle time, HyperSight CBCT produced ITV volumes and HU values comparable to 4DCT at small respiratory amplitudes (≤10 mm), with Dice similarity coefficients exceeding 0.8 and dose calculations passing 1%/1 mm gamma analysis. At larger amplitudes (≥15 mm), ITVs were underestimated to 28-40% of the reference values, with Dice coefficients falling below 0.45 and increasing image distortion. Longer cycle times (8s) and irregular breathing patterns (hysteresis) further reduced trajectory visualization, while more uniform motion (sinusoidal) improved trajectory coverage but still showed image deformation at 15mm amplitude. Repeat acquisitions demonstrated varying image representations at different scan initiation times, particularly with the longer 8-second cycle. Slow-scan CBCT achieved close agreement with 4DCT MIP and AIP in ITV volumes and HU values. CONCLUSIONS: HyperSight CBCT is feasible for OART in patients with limited respiratory motion, regular breathing pattern and short breathing cycle time. For larger excursions, slow scans or motion management strategies may be required to ensure accurate target delineation and dose calculation.