ESTRO 2023 - Abstract Book

S1401

Digital Posters

ESTRO 2023

reconstructed CBCT is therefore essential to prevent incorrect couch shifts or geographic irradiation errors with respect to target volumes and OARs. The aim of this study was to determine the influence of different CBCT reconstruction algorithms on bone match results. Materials and Methods To evaluate the influence of difference CBCT reconstruction algorithms on bone match results, two types of new_CBCT images were created: an iterative reconstruction and a synthetic CT (sCT) generation with deep-learning. Iterative reconstruction was performed with Framework for Reconstruction and Simulation of Cone beam images (FRESCO) V21.1.0 and the sCT generation was performed with ADMIRE research V3.37.0 Male-Female Pelvis model v.2.0 (Elekta AB, Sweden). Images of 12 cervical cancer patients were included. Patients were treated according to EMBRACEII. Two CBCT images of two different fractions and one pCT with contrast were selected per patient. pCT and CBCT images were acquired on a Philips big bore CT (Philips, Netherlands) and XVI system (Elekta AB,Sweden), respectively. Bone match between pCT and CBCTs was performed in Raystation V.10b. Differences in match results of the XVI_CBCT, FRESCO and ADMIRE images were calculated. To test the results on significance, a Wilcoxon Singed Rank test was used with p<0.05. Results The image quality was improved in ADMIRE_CBCT (Figure 1). PSNR in the bladder was largest in ADMIRE_CBCT compared to FRESCO_CBCT and XVI_CBCT, with 40±3 dB, 35±2 dB and 35±4 dB (mean±std), respectively. There was a significant difference in posterior-anterior direction between the XVI_CBCT match and ADMIRE_CBCT match, -0.93±0.16 mm (mean±STD) (p=0.000) (Figure 2C). No significant difference was found between XVI_CBCT match and FRESCO_CBCT match, -0.03±0.17 mm (p=0.797). In left-right and superior-inferior direction, no clinically relevant differences were found (mean <0.03 mm). Also no clinically relevant differences were found in rotations (mean <0.2 degree).