ESTRO 2025 - Abstract Book

S3037

Physics - Image acquisition and processing

ESTRO 2025

Purpose/Objective: While the MRI-only workflow in radiotherapy offers well-known advantages, its clinical implementation requires validated procedures tailored to the specific work environment. This study provides a step-by-step assessment of implementing the MRI-only workflow for pelvic treatments in a clinical setting using multiple linear accelerators, treatment plans, MR and CT scanners. Focus is given to three key phases: pretreatment imaging, treatment plan, and treatment delivery. Material/Methods: A total of 55 patients receiving pelvic radiotherapy were included in this study. Patients, set-up supine using standard immobilization systems, underwent both CT and MRI scans with compatible markers aligned to reference tattoos. Radiation oncologists delineated targets first on MRI images alone, then on CT-MRI fused images following routine clinical practice. Reference treatment plans (TP REF ) were generated using simulation CT images, while synthetic CT (sCT) based plans (TP sCT ) were obtained by recalculating TP REF using MRI-based sCT images with bulk density assignments. For treatment verification, daily CBCT with an extended FOV was acquired and matched online to the reference CT. Offline MRI-CBCT registrations were also performed recording displacements in the x, y, z directions for both modalities. The workflow is illustrated in Figure 1. The delineated MRI-only targets were compared to CT-MRI fused targets using volume mean differences and Hausdorff Distance (HD). Dosimetric agreement between TP sCT and TP REF was evaluated through dose-volume histograms (DVH) and gamma analysis. Patient positioning accuracy was accessed by comparing CBCT-to-MRI with the corresponding CT-CBCT matches. Statistical analysis were performed using SPSS software, with paired t-tests or Wilkinson signed-rank tests applied as appropriate.

Figure 1 Standard versus MRI-only workflow assessment

Results: The MRI-only workflow was evaluated for 22 rectal, 10 prostate, 13 cervical, and 10 gynecological cancer patients. MRI-only targets were significantly smaller than corresponding that the MR-CT fused volumes ( p< 0.05 ). The HD median value was 5.2 mm for target smaller than 50 cm 3 and 15.1 mm for larger targets. Target and OARs DVHs from TP REF and TP sCT showed no significant differences ( p>0.05 ); results for targets are detailed in Table 1. The