S2053
Physics - Image acquisition and processing
ESTRO 2026
Digital Poster Highlight 1163
With fractional Blood Volume (fBV), pseudo-diffusion coefficient (D*) and signal s(b) from multi b-values DW- MR images.Then, a repeatability coefficient (RC (%)), measuring the stability of qMRI measurements across different measures, was calculated for each patient according to [2], a RC below 68% indicating good repeatability of the ADC on FHs [1]. This RC was calculated for ADC between the simulation and the 1st session of irradiation, and among all irradiation session (from 1st to last one). Results: Across all patients FHs, mean RC of 23.9% and 18.6% (Figure1) were found for ADC between simulation and 1st session, and from 1st to last session. These RC results were below the threshold of 68%, indicating good repeatability of the ADC measurements on FHs in one month interval or during SBRT treatment.
Slicer IMPACT-Synth: An Open 3DSlicer plugin for Synthetic CT Generation and Integrated Quality Assurance in Radiotherapy Cédric Hémon, Valentin Boussot, Jean-Claude Nunes, Jean-Louis Dillenseger, Anais Barateau, Renaud de Crevoisier, Caroline Lafond Univ. Rennes, CLCC Eugène Marquis, INSERM, LTSI - UMR 1099, Rennes, France Purpose/Objective: With the rapid growth of AI applications in radiotherapy, generating high-quality synthetic images has become increasingly important for improving treatment planning workflows. IMPACT-Synth is the first fully open-source framework for synthetic CT (sCT) generation from MRI and CBCT for whole-body. In this work, we introduce a dedicated 3D Slicer plugin that interfaces with the IMPACT-Synth framework and makes it usable within a widely adopted open-source platform for routine clinical and research workflows in radiotherapy. Material/Methods: The plugin integrates state-of-the-art deep learning models from the SynthRAD2025 Challenge [1], ranked third in both MRI-to-CT and CBCT-to-CT tasks. Powered by KonfAI [2], a modular and configurable framework for medical imaging AI, it ensures fast inference, flexible deployment, and seamless integration within 3D Slicer.Slicer IMPACT-Synth supports RT DICOM series, NIfTI, and MHA formats, requiring no specific preprocessing. MR, CBCT, or planning CT data can be directly loaded from PACS or local folders for sCT generation.Beyond synthesis, the plugin exposes the Quality Assurance (QA) module of IMPACT-Synth, enabling anatomical conformity assessment and uncertainty quantification. In its current version, the QA process relies on:Automatic segmentation of both the sCT and the daily image.Quantitative comparison by structure, using Dice and surface distance metrics.Visual inspection through synchronized multi- view display.Uncertainty assessment through maps computed via ensemble and test-time augmentation strategies, capturing both aleatoric and epistemic uncertainties.Uncertainty maps allow users to visually identify regions of low confidence and propagate these uncertainties to dose calculations, following the predictive modeling approach described in [3]. The upcoming version will integrate supervised and unsupervised generation approaches to assess anatomical consistency. Supervised models may yield high fidelity but create artifacts, while unsupervised ones offer greater robustness. Results: Quantitative results on the SynthRAD2025 dataset show consistent performance across anatomical
Conclusion: Good repeatability was demonstrated for the ADC within FHs across patient population. This study demonstrated the reliability of new qMRI sequences on MR-Linac for assessing in-treatment biomarkers with the aim of enabling accurate real-time assessment of tumour progression. References: [1] Michoux et al. Repeatability and reproducibility of ADC measurements: a prospective multicenter whole- body-MRI study. Eur Radiol. 2021[2] Boss et al. The QIBA Profile for Diffusion-Weighted MRI: Apparent Diffusion Coefficient as a Quantitative Imaging Biomarker. Radiol. 2024[3] Mesny et al. Towards tumour hypoxia imaging: Incorporating relative oxygen extraction fraction mapping of prostate with multi-parametric quantitative MRI on a 1.5T MR-linac. J Med Imag and Rad Onco. 2024[4] Hompland et al. Combined MR Imaging of Oxygen Consumption and Supply Reveals Tumor Hypoxia and Aggressiveness in Prostate Cancer Patients. Cancer Res. 2018 Keywords: MR-Linac, Repeatability, Quantitative MRI
Made with FlippingBook - Share PDF online