ESTRO 2025 - Abstract Book

S3002

Physics - Image acquisition and processing

ESTRO 2025

[ 4 ] Barral et al. Magn Reson Med 2010 64 , 1057–1067. [ 5 ] In den Kleef et al. Magn Reson Med 1987 5 , 513–524. [ 6 ] Deoni et al. Magn Reson Med 2003 49 , 515–526. [ 7 ] Nguyen et al. Magn Reson Med 2017 78 , 518–526. [ 8 ] Baudrexel et al. Magn Reson Med 2018 79 , 3082–3092. [ 9 ] Shukla-Dave et al. J Magn Reson Imaging 2019 49 , e101–e121.

2087

Digital Poster Evaluation of HyperSight cone-beam CT imaging for contouring and dose recalculation in head-and-neck cancer patients Colien Hazelaar 1 , Vicki Trier Taasti 1 , Frederik Wesseling 1 , Anne van Engelen 1 , David Hofstede 1 , Nikolina Birimac 1 , Maud Daniels 1 , Laura Ligter 1 , Eline Palmen 1 , Jens de Wit 1 , Sean R.H. Davidson 2 , Marta Bogowicz 1 , Wouter van Elmpt 1 1 Department of Radiation Oncology (Maastro), GROW Research Institute for Oncology and Reproduction, Maastricht University Medical Centre+, Maastricht, Netherlands. 2 Clinical Operations, Varian Medical Systems, Palo Alto, USA Purpose/Objective: To evaluate the quality of CBCT scans acquired with an advanced CBCT imaging system of head-and-neck cancer patients for contouring and dose recalculation. Material/Methods: For seven head-and-neck cancer patients, extended CBCT scans were acquired on a Halcyon system with HyperSight imaging (v4.0, Varian) in a clinical imaging trial (NCT05524454). Scans were reconstructed twice, using the filtered back-projection algorithm (HS-FDK) and advanced iterative reconstruction including Acuros scatter correction and metal artifact reduction (HS-MAR). Image quality was evaluated quantitatively and qualitatively and compared to the planning CT (pCT) scan (acquired 7-28 days earlier). HyperSight scans were rigidly registered (6 DoF) to the pCT, and spherically-shaped regions-of interest (0.1-0.8 cm 3 ) were placed in homogeneous areas of muscle, fat, and submandibular glands. Mean CT number in the regions-of-interest and standard deviation (SD) were calculated. For patients without large anatomical deformations resulting in inaccurate image registration, dose recalculation was performed on the HS MAR scans, applying a dedicated CBCT CT-number-to-mass-density conversion curve, in Eclipse (Acuros v16.1). DVH parameters were evaluated on both rigidly copied (from pCT to HS-MAR) and manually edited deformably copied contours. Additionally, seven experienced observers qualitatively scored their confidence in contouring specific OARs on HS-MAR and pCT scans using a 5-point Likert scale. Results: For muscle and submandibular glands, noise increased from pCT to HS-MAR/HS-FDK: e.g., for muscle, median SD was 6.6 HU (pCT), 9.8 HU (HS-MAR), and 11.2 HU (HS-FDK). For fat, median SD was 8.9 HU for pCT and 7.1/7.6 HU for HS-MAR/HS-FDK. Differences in DVH-parameter values between HS-MAR and pCT were ≤1.7% (1.1 Gy) for rigidly copied OARs (Figure 1). Target DVH-parameters showed larger differences due to changes in air near the target. For propagated/recontoured OARs, differences were up to 10% (6.5 Gy) due to anatomical and contouring variations.