S2491
Physics - Radiomics, functional and biological imaging, and outcome prediction
ESTRO 2026
Material/Methods: The proposed framework, summarized in Figure 1, was validated on an anatomically realistic XCAT [2] phantom with detailed cerebral and vascular structures. Tissue-specific CBF and CMRO ₂ maps were assigned to generate anatomically and physiologically realistic ground-truth profiles. In particular, heterogeneous CBF patterns from healthy subjects [3] were transferred through affine resampling using weighted k-nearest neighbours in a normalized feature space. Non-blood susceptibility ( χ nb) was computed analytically to preserve QSM consistency, while total susceptibility ( χ ) was derived from venous and non-blood sources [1].The inverse problem for physiological parameter recovery was solved region- wise using a quadratic programming approach, with Tikhonov regularization, designed to improve numerical stability and CMRO ₂ and χ nb estimation, without straight-sinus calibration required in previous approach [1]. Estimation errors were assessed by tissue-based and global RMSE against ground-truth maps. A realistic QSM phantom from the QSM Challenge 2.0 [4] was used for validation under realistic conditions, with literature-derived uniform CBF values map. Heterogeneous voxels were agglomerated into homogeneous clusters, each assigned to a specific anatomical region. Estimated CMRO ₂ values were compared with reported physiological ranges for gray matter, thalamus, putamen and globus pallidus [5]. Tissue pO ₂ was subsequently derived by estimating the oxygen extraction fraction (OEF) and SvO2 through analytical methods via oxyhemoglobin dissociation curve [6].
for quantitative results.
Conclusion: This work confirms that the proposed QSM-based framework can reliably recover physiologically meaningful CMRO ₂ and pO2 maps in realistic conditions. Future work will integrate patient-specific perfusion maps to enable personalized, non-invasive identification of hypoxic regions towards biologically targeted radiotherapy. References: [1] Zhang J, et al. Magn Reson Med. 2018, doi: 10.1002/mrm.26657.[2] Segars WP, Med Phys, 2010 doi: 10.1118/1.3480985.[3] Meng,Z; et al. 2020, doi: 10.6084/m9.figshare.28093595.v1[4] Marques JP, et al. 2020 QSM doi:10.1101/2020.09.29.316836[5] Song H, 2023, doi: 10.1016/j.zemedi.2023.07.004. [6] Dahl RH, et al. Exp Physiol. 2020 doi: 10.1113/EP088615. Keywords: MRI, susceptibility, hypoxia Volumetric dose and image-based analysis of the vaginal wall for predicting vaginal stenosis after cervical cancer brachytherapy C.J.A. (Maaike) Romme 1,2 , Jan Wiersma 3 , Henrike Westerveld 4 , Peter A.N. Bosman 5 , Luc R.C.W. van Lonkhuijzen 1,2 , Lukas J.A. Stalpers 3 , Tanja Alderliesten 6 1 Department of Gynaecology and Obstetrics, Amsterdam University Medical Centers, location Vrije Universiteit Amsterdam, Amsterdam, Netherlands. 2 Treatment and Quality of Life, Cancer Center Digital Poster 4726 Amsterdam, Amsterdam, Netherlands. 3 Department of Radiation Oncology, Amsterdam University Medical Centers, location University of Amsterdam, Amsterdam, Netherlands. 4 Department of Radiotherapy, Erasmus Medical Center Cancer Institute, University Medical Center Rotterdam, Rotterdam, Netherlands. 5 Evolutionary Intelligence Group, Centrum Wiskunde & Informatica, Amsterdam, Netherlands. 6 Department of Radiation Oncology, Leiden University Medical Center, Leiden, Netherlands Purpose/Objective: Patients with cervical cancer receiving brachytherapy (BT) may develop vaginal stenosis, which can significantly affect quality of life. Identifying high-risk patients early on could guide counselling and follow-
Results: Both digital and realistic validations demonstrated stable convergence and physiologically consistent CMRO ₂ reconstructions. In the XCAT phantom, estimated CMRO ₂ values matched ground truth (global RMSE of 1.2 10-12 µmol/100 g/min; regional RMSE<10- 9 µmol/100 g/min), confirming numerical robustness. In the control case, CMRO ₂ and pO2 estimates remained within reported physiological ranges for gray and white matter (pO2: 20-50 mmHg); see Table 1
Made with FlippingBook - Share PDF online