ESTRO 2025 - Abstract Book

S3007

Physics - Image acquisition and processing

ESTRO 2025

Material/Methods: We performed phantom scans at a clinically available PCD-CT scanner (Siemens NAEOTOM Alpha) at 120 kVp, using clinical radiology protocols. Different virtual mono-energetic images (VMI), electron density (ED) images, with iterative metal artefact reduction (iMAR), as well as iodine maps were reconstructed. We calibrated the conversion of HU to ED for different VMI using a CIRS electron density phantom. ED images directly provide the material parameter of interest for dose calculation. The Gammex MultiEnergy phantom was scanned with metal inserts (aluminum, titanium, steel), iodine concentration, and calcium concentration, in blood-mimicking tissue-equivalent inserts. Results: HU-to-ED calibration changes for different VMI due to the variation in relative importance of photoelectric effect (PE) and Compton scattering (CS) with photon energy. The calibration obtained for 70 keV VMI (VMI70) matches a conventional CT calibration using 120 kVp, while higher energies show lower HU values. Figure 1 compares HU values in VMI70 vs. VMI190, as a surrogate for the [PE, CS] space, demonstrating material separation based on energy-dependent interactions. On VMI70 we found low normalized dispersion-to-separation ratio (DtS) between iodine 10 and 15 mg/ml (DtS = 0.22), and between iodine 5 mg/ml and calcium 5 mg/ml (DtS = 0.18). Since iodine maps directly quantify iodine concentration, which we verified to be linear (r² = 0.99) with different iodine inserts, contrast liquid can be detected and corrected in planning-CT scans, sparing the patient the need to perform two CT scans (with and without contrast). As HU in PCD-CT have a large dynamic range, we could also detect very dense materials, identified using information from both VMI70 and VMI190. As a proof of principle, we compared the dose calculation for a patient where all dense materials are overridden as titanium (standard practice when implant material is unknown) with dose calculation performed on an ED map where we could separately assign materials to different implants, observing dose differences of up to 10% close to the denser implants (Figure 2).