ESTRO 2025 - Abstract Book

S3126

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2025

Purpose/Objective: To directly use HyperSight Cone-Beam CT (CBCT) for re-planning in offline adaptive radiotherapy, it is essential to quantify its imaging performance concerning various tasks in treatment planning. This study aims to quantitatively evaluate the impact of Hounsfield unit (HU) correction of CBCT on the deformable image registration between the planning CT and CBCT. Material/Methods: HyperSight CBCTs from a total number of thirty sessions from five Head-and-Neck (HN) patients and a total number of eighteen sessions from three prostate patients treated with Ethos were included in this study. We adjusted the HU values of CBCT based on pre-established HU-density relationships from simulation CT scans 1 . Sixteen organs-at risk (OARs) were automatically segmented on the planning CT (pCT) of HN patients, including Brainstem, Parotids (2), Submandibular glands (2), Oral Cavity, Larynx, Thyroid cartilage, Brachial Plexus (2), Spinal Cord, Mandible, Pterygoid muscles (2), and Masseter muscles (2.) For prostate patients, five OARs were segmented, including Prostate: Rectum, Bladder, Penile bulb, Urethra, and Femurs (2.) For each session, we performed two pCT-to-CBCT deformable image registrations (DIRs), one with the original CBCT (DIR-org) and one with the HU-adjusted CBCT (DIR-adj.) We used an intensity-based method in commercial software to perform DIRs and contours were propagated from the planning CT to session CBCTs using the DIRs. To assess the difference between two sets of propagated contours from DIR-org and DIR-adj, we quantified the similarity of the pair of contours with Dice similarity coefficient (DSC) and Hausdorff distance (HD.)

Results: