ESTRO 2025 - Abstract Book

S3006

Physics - Image acquisition and processing

ESTRO 2025

Figure 2. Bar graph for the agreement among expert delineation. Each bar represents a different type of computed tomography (CT), with three data groups: EID-CT, PCD-CT 140 keV, and PCD-CT 50 keV + 140 keV. The p values provided indicate the statistical significance of the observed differences.

Conclusion: The study established the feasibility of PCD-CT for radiotherapy planning, demonstrating that PCD-CT-based delineation provides results comparable to conventional EID-CT, with the added advantage of reduced radiation dose and improved diagnostic image quality. Significant improvements in delineating specific structures, such as the lips and oral cavity, were observed when combining multiple energy levels in PCD-CT. These findings highlight the potential of PCD-CT to enhance segmentation accuracy in radiotherapy and support its integration into clinical workflows.

Keywords: photon counting CT, segmentation, head and neck

2176

Digital Poster First implementation of photon counting CT in radiotherapy dose calculation for material distinction and identification Serena Psoroulas 1 , Mohammad Varasteh Anvar 1 , Davide Cester 2 , Enkelejda Lamaj 1 , Lotte Wilke 1 , Patrick Wohlfahrt 3 , Hatem Alkadhi 2 , Bettina Zanini 2 , Niccolo Bertini 4 , Hubert Gabrys 1 , Matthias Guckenberger 1 , Panagiotis Balermpas 1 , Stephanie Tanadini-Lang 1 1 Radiation Oncology, University Hospital Zurich, Zurich, Switzerland. 2 Diagnostics and Interventional Radiology, University Hospital Zurich, Zurich, Switzerland. 3 R&D, Siemens Healthineers, Forchheim, Germany. 4 Radiation Oncology, Careggi University Hospital, Florence, Italy Purpose/Objective: Photon counting detector CT (PCD-CT) is an innovative imaging modality which we plan to integrate in our radiotherapy workflow. We investigated different image reconstruction options and their CT number (so-called Hounsfield Units, HU) accuracy for different materials such as contrast agents and metals.