ESTRO 2025 - Abstract Book

S3654

Physics - Quality assurance and auditing

ESTRO 2025

2989

Digital Poster Optimizing dose in CBCT protocols: A phantom and clinical study of the iCBCT algorithm in head and neck region Ana M Sania, Carme Ares, Gabriela Antelo, Izaskun Valduvieco, Jordi Tarrats, Artur Latorre-Musoll Department of Radiation Oncology (ICAMS), Hospital Clinic Barcelona, Barcelona, Spain Purpose/Objective: Varian presents the iCBCT as an iterative reconstruction algorithm for cone-beam computed tomography (CBCT) that enhances image quality through scatter correction and statistical reconstruction. Given the increasing use of CBCT in daily image-guided radiation therapy (IGRT) protocols, their optimization becomes more important. The aim of this study is to explore the effectiveness of this novel algorithm to achieve dose-image optimization in the head and neck (H&N) region for adult patients. Material/Methods: In a preliminary study using the Catphan phantom (model 604), acquisitions were performed at 100 kVp and varying exposure levels, ranging from 10 mAs (the minimum allowed) to 150 mAs (default manufacturer protocol for H&N region), to characterize image noise in the iCBCT compared to the conventional algorithm, Feldkamp–Davis–Kress algorithm-based CBCT (FDK-CBCT). In a second study, we assessed the accuracy of the on-line bony-matching automatic registration using both algorithms. We repeated the same acquisitions over an anthropomorphic RANDO phantom using a set of known shifts including translations and rotations of up to 1 cm and 1º. Finally, a study was conducted on clinical practice involving H&N region for adult patients. The mAs was progressively decreased in each treatment session, starting from the initial 150 mAs to the level deemed acceptable by the radiation oncologist. Results: Figure 1 shows that the iCBCT algorithm reduces image noise by an average of 38%, regardless of the exposure level. Approximately, while maintaining the default manufacturer protocol for H&N region image noise, the iCBCT algorithm achieves a mAs reduction of 63%. In the analysis of registration accuracy, errors were found to be below 0.05 mm and 0.1º in all reconstructions. Therefore, in the dose optimization process, registration accuracy is not a limiting factor. The clinical study led us to create two distinct protocols (Figure 2). For brain tumors, the minimum value of 10 mAs could be reached, as suggested by the registration accuracy study. However, in head and neck (H&N) regions including shoulders, exposure could only be reduced to 50 mAs due to a significant loss of image quality in the shoulder area (truncated images are a known limiting factor in iterative algorithms).