S2052
Physics - Image acquisition and processing
ESTRO 2026
MR-only prostate radiotherapy. Thomas C, Dregely I, Oksuz I, Guerrero Urbano T, Greener T, King AP,
Barrington SF. BJR Open. 2024 Jun 3;6(1) Keywords: Synthetic CT, MR-only, Prostate
Digital Poster 1152 Repeatability of quantitative MRI sequences on a 1.5T MR-Linac Valentin Septiers 1,2 , Joséphine Colineaux 1 , Léo Le Bozec 1 , Jennifer Le Guévelou 1 , Manon Baty 1 , Julie Faucheux 1 , Christophe Fusel 1 , Clément Guibert 1 , Alice Le Jeanne 1 , Lucie Pelhate 1 , Maxime Yon 1 , Hervé Saint- Jalmes 1 , Renaud De Crevoisier 1 , Maria A Zuluaga 2 , Oscar Acosta 1 , Anaïs Barateau 1 1 CLCC Eugène Marquis – INSERM - LTSI - UMR 1099, LTSI, University of Rennes, Rennes, France. 2 Data science department, EURECOM, Biot, France
Purpose/Objective: The quantification of quantitative MRI (qMRI)
biomarkers is crucial for anticipating tumour response, adapting irradiation doses and enabling personalised monitoring during and after treatment. The recent development of 1.5T MR-Linac, offers a unique opportunity to acquire longitudinal and qMRI data throughout RT. However, the reliability of qMRI biomarkers such as the Apparent Diffusion Coefficient (ADC) is always questioned on new devices to consider whether or not a change in ADC can be considered as a real change in tumour biology [1,2]. This study aims to evaluate the repeatability of qMRI sequences over time during MRI-guided RT. Material/Methods: In this study, 13 patients with localised prostate cancer were treated with stereotactic body RT (SBRT) (Frontline 5 × 8Gy) on an Elekta Unity 1.5T MR-Linac. A simulation was performed one month before the treatment's start. At each session (simulation included), optimised qMRI sequences were acquired: T2w and DW (Diffusion Weighted) images based on multi-b values (0;5;20;40;80;200;500 s/mm2) and fat suppression (SPAIR), based on established MR- Linacprotocolsfrom another center [3]. Femoral heads (FHs) were delineated by an expert radio-oncologist on T2w images and contours propagated on co-registered DW images. This location was chosen for assessing the repeatability of qMRI sequences as no changes were expected before and through treatment. The ADC within FHs was extracted using an IVIM (IntraVoxel Incoherent Motion) model (Equation (1)) by performing a pixel-wise fitting with signal from multi b- values DW-MR sequence [4]:
60Gy planning target volume (PTV60), rectum and bladder median dose (D50%) errors and standard deviations were (0.25% +/- 0.38%), (0.01% +/- 0.08%) and (0.01% +/- 0.11%) respectively. Mean 2%/2mm and 1%/1mm gamma pass rates within the high dose area (50% isodose) were 100.0% +/- 0.0% and 99.98% +/- 0.1% respectively.
Conclusion: The sCT model appears generalisable to images acquired using a different scanner and imaging protocol, despite significant differences in acquisition parameters and image pixel values. This suggests the model is robust and could be used across multiple centres subject to local validation, enabling MR-only radiotherapy to be more widely adopted. References: [1] Image-Based Deep Learning Enables the Reduction of GastroIntestinal Toxicity in Pelvic Radiotherapy, Thomas, C. (Author). 1 Oct 2023, Student thesis: Doctoral Thesis › Doctor of Philosophy [2] Effect of synthetic CT on dose-derived toxicity predictors for
Made with FlippingBook - Share PDF online