S2048
Physics - Image acquisition and processing
ESTRO 2026
Induced Signal Misallocation Artifacts in Two-Point Fat- Water Chemical Shift MRI. Magn Reson Med Off J Soc Magn Reson Med Soc Magn Reson Med 2015; 73: 1926–1931. Keywords: MR-visible material, MR-only, Microcapsules
UV-curable resin as matrix material and cured in bulk. Imaging was performed on a 1.5 T MR scanner (Ingenia Ambition X, Philips, Eindhoven, The Netherlands; v5.7). T1 relaxation times were measured using an inversion recovery sequence run multiple times with varying inversion times from 25 to 4000 ms. T2 decay times were evaluated using a multi echo turbo spin sequence with echo times ranging from 10 to 160 ms. Mean intensity values were extracted and used for fitting the respective relaxation and decay functions. For the T1 data a biexponential fit with both T1 times set to the predetermined values of pure paraffin oil and water was performed. A DIXON sequence was used to separately visualize the water and oil signal from the material. The uncertainty of the fitting parameters T1 and T2 reflect the noise and heterogeneity of the sample rather than repeated measurements. Results: Our novel material showed clear positive contrast in DIXON- (Fig.1), inversion recovery-, and turbo spin- sequences. Fitting a biexponential model with both T1 times set to the predetermined values of paraffin oil and water resulted in an R2 of 0.991 and a weighting factor of 0.51 for the paraffin oil component. The determined T2 time of the material was 129±13 ms (R2=0.989), pure paraffin oil had a T2 time of 118±2 ms (R2=0.996).
Digital Poster Highlight 1016
Automated, rapid QC of commercial pelvis synthetic CT using independent in-house GAN model and TotalSegmentator Akos Gulyban 1 , Kevin Brou Boni 1 , Zelda Paquier 1 , Manuela Burghelea 1 , Nicolas Jullian 2 , Nick Reynaert 1 1 Medical Physics, Hôpital Universitaire de Bruxelles (HUB), Institut Jules Bordet, Brussels, Belgium. 2 Radiation Oncology, Hôpital Universitaire de Bruxelles (HUB), Institut Jules Bordet, Brussels, Belgium
Purpose/Objective: The implementation of MRI-only workflows in
radiotherapy planning requires reliable synthetic CT (sCT) generation and robust quality control (QC) to ensure geometric and dosimetric accuracy. This study presents a fully automated QC pipeline for commercial pelvis synthetic CTs via independent generative adversarial network (GANs) and the integration of TotalSegmentator[1] framework for quantitative evaluation. Material/Methods: Two GAN-based pelvis synthetic CT models were used: a vendor-provided (Siemens, [2], “sCT”) model (externally trained) for clinical routine and an in-house trained AugCGAN model (“ksCT”, [3], N=20) for independent verification. Both models used identical Dixon in-phase and out-of-phase MR inputs of the pelvis region.The automated QC pipeline was designed to operate from a single DICOM export, generating sCT and ksCT (runtime: Siemens sCT <3 min; ksCT <1 min). Each output was processed by TotalSegmentator in both high-resolution (HR, 1.5 mm) and low-resolution (LR, 3 mm) to segment thirteen (7 bone and 8 muscle, soft tissue) anatomical regions of interest (ROIs). For each ROI, the volume and mean electron density (ED) were calculated using the corresponding CT specific ED-HU table and a comparison of HR and LR results were performed using paired t-test (with p<0.05 significance level). Based on the validation cohort (N=21), the volume-ED results were convoluted to ellipsoids using 2.5 SD intervals to define acceptance limits (“circle of trust”) for each ROI in order to detect anomalies such as signal loss, prostheses, or reconstruction errors. Results: Both sCT models demonstrated high anatomical fidelity and homogeneity, particularly in bony
Conclusion: We were able to successfully prepare a solid composite material which produces well visible positive contrast in MRI. Therefore, paraffin oil filled microcapsules are a suitable additive to render polymers MRI-visible. T1 time analysis revealed that our fabrication process also incorporates water into
the material. References:
[1] Axelson DE, Kantzas A, Nauerth A. 1H Magnetic resonance imaging of rigid polymeric solids. Solid State Nucl Magn Reson 1996; 6: 309–321.[2] Chandarana H, Wang H, Tijssen RHN, et al. Emerging Role of MRI in Radiation Therapy. J Magn Reson Imaging JMRI 2018; 48: 1468–1478.[3] Tanderup K, Viswanathan A, Kirisits C, et al. MRI-guided brachytherapy. Semin Radiat Oncol 2014; 24: 181– 191.[4] Rahimi MS, Holmes JH, Wang K, et al. Flow-
Made with FlippingBook - Share PDF online