S2497
Physics - Radiomics, functional and biological imaging, and outcome prediction
ESTRO 2026
volunteers underwent repeated pelvic IVIM on both a 1.5 T Unity® MR-Linac (Philips-Elekta) and a 1.5 T diagnostic MRI (Siemens) to evaluate repeatability and reproducibility of pelvic organs. Finally, rectal tumors were assessed in sixteen patients treated on the MR- Linac. Test–retest variability was assessed using short- term (two acquisitions on the same day) and long-term (7–15 days apart) measurements. Regions of interest were manually delineated on high b-value images using MIM® software. IVIM-derived D and f maps were computed, and the repeatability coefficient (RC) was calculated following QIBA recommendations. Data were reported as medians. Several parameters were varied to assess their impact: for each healthy organ, a fixed ROI and a half-sized ROI were analyzed to assess volume effects, and two sequences were evaluated (the Unity consortium reference and a higher- resolution one). Results: Phantom measurements met QIBA performance standards, with short-term and long-term repeatability showed in Table 1. For the patients and healthy volunteers, as expected, perfusion-related parameters exhibited greater variability than diffusion. Repeatability values for rectal tumors and prostate were consistent with those previously reported in the literature. The segmentation volume did not affect the results, and the study sequence has a better RC than the reference one. Reproducibility between the MR- Linac and the diagnostic MRI (Table 2) was acceptable for diffusion but remained limited for perfusion. Treatment-related changes exceeded the repeatability threshold in 9/16 rectal tumors for IVIM-D (>10.7%) and 1/16 for IVIM-f (>20.8%).
Fig.2-Trend of GTV-ADC during treatment for R and NR. Conclusion: Serial ADC measurements acquired on a 0.35 T MR- Linac potentially demonstrate the ability to discriminate between ERI-defined responders and non-responders throughout most of a 25-fraction course. ADC dynamics—characterized by early, mid- treatment peaks and an overall on-treatment rise— support the role of qMRI as a practical biomarker to refine response assessment during treatment and potentially enhance ERI-driven adaptive workflows with patient-specific online biomarkers. Prospective validation and threshold optimization are warranted to translate these findings into decision rules for online adaptation. References: 1. Fiorino, C. et al. Early regression index (ERI) on MR images as response predictor in esophageal cancer treated with neoadjuvant chemo-radiotherapy: Interim analysis of the prospective ESCAPE trial. Radiother. Oncol. J. Eur. Soc. Ther. Radiol. Oncol.194, 110160 (2024).2. Chiloiro, G. et al. THUNDER 2: THeragnostic Utilities for Neoplastic DisEases of the Rectum by MRI guided radiotherapy. BMC Cancer22, (2022). Keywords: MR-Linac, qMRI, biomarkers-guided adaptive RT. Digital Poster Highlight 5127 Ensuring reliability of DW-MRI biomarkers : repeatability and reproducibility in pelvic organs on a 1.5 T MR-Linac Madeline Michel 1 , Zelda Paquier 2 , Philippe Martinive 1 , Robbe Van den Begin 1 1 Radiation oncology, Institut Jules Bordet, Brussels, Belgium. 2 Medical Physics, Institut Jules Bordet, Brussels, Belgium Purpose/Objective: Diffusion-weighted MRI (DW-MRI) biomarkers are increasingly investigated to guide adaptive MR-guided radiotherapy (MRgRT) and predict treatment response for rectal cancer. However, their clinical translation critically depends on robust repeatability and reproducibility across systems. To support reliable multicenter implementation, we investigated the repeatability and reproducibility of intravoxel incoherent motion (IVIM) parameters (diffusion (IVIM- D) and perfusion fraction (IVIM-f)), for pelvic organs on a 1.5 T MR-Linac and a 1.5 T diagnostic MRI. Material/Methods: Initial acquisitions were performed in vitro using the QIBA (Quantitative Imaging Biomarkers Alliance) diffusion phantom. Subsequently, thirteen healthy
Conclusion: IVIM-D was repeatable with acceptable inter-system reproducibility, whereas IVIM-f was highly variable;
Made with FlippingBook - Share PDF online