ESTRO 2025 - Abstract Book

S2991

Physics - Image acquisition and processing

ESTRO 2025

1907

Digital Poster Optimizing MRI Coil Setups for Enhanced DTI in Radiotherapy with Immobilization Devices: Towards Clinical Integration of White Matter Tracts Sophie Bockel 1,2 , Lucie Calmels 1 , Naila Boudiaf 3 , Noura Sellami 1 , Roger Sun 1,2 , Camilla Satragno 1,2 , Cédric Yuste 1 , Eliot Nicolas 1 , Virginie Chiesa 1 , Najma Douir 1 , Tiphanie Anou 1 , Noemie Delivert 1 , Virginie Le Merlus 1 , Christophe Geran 1 , Cécile Bardoulat 1 , Elodie Barthelemy 1 , Ines Yahyaoui 1 , Frédéric Dhermain 1 , Eric Deutsch 1,2 , Charlotte Robert 1,2 1 Radiotherapy, Gustave Roussy, VILLEJUIF, France. 2 INSERM U1030, Gustave Roussy, VILLEJUIF, France. 3 MR Collaboration, Siemens Healthineers, ERLANGEN, Germany Purpose/Objective: Advances in imaging are crucial for brain tumor radiotherapy (RT), with a focus on preserving neurocognitive brain structures. Diffusion Tensor Imaging (DTI) aids RT planning by mapping white matter tracts, but high-quality DTI acquisition in RT immobilization setups is challenging. This study evaluated MRI coil configurations under RT constraints to determine the optimal setup based on signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) metrics. Material/Methods: Experiments were conducted using a 3T MRI scanner (MAGNETOM Vida, Siemens Healthineers, Erlangen, Germany) with a CIVCO3400 RT table. Five coil configurations were tested on five healthy subjects (see Figure 1): • Reference (Diagnostic-like image quality) : 20-channel head coil without RT immobilization. • Tests with thermoplastic mask : 1. Two 18-channel UltraFlex (18UFL) coils around the head. 2. One 18UFL coil atop the head. 3. One 4-channel small Flex (4FS) coil behind and one 18-channel Body coil (18BC) atop the head. 4. One 4FS coil behind and one 18UFL atop the head. 5.

DTI images were acquired in axial plane, Field Of View = 230 mm x 230 mm x 140 mm; Matrix= 184 x 184 x 56; Voxel= 1.25 mm x 1.25 mm; Slice thickness= 2.5 mm; TR/TE= 4100/76 ms; Flip Angle= 90°; b-values= 0 and 1000 s/mm2, 64 diffusion directions, acceleration factor of 4, and with anterior-posterior phase encoding direction. Preprocessing of raw data included Marchenko-Pastur PCA denoising, correction of susceptibility-induced distortion and eddy current artifacts (Topup and Eddy FSL[1]). Synthesized b0 for diffusion distortion correction was used (Synb0-DISCO[2]) as reverse phase-encoding scans were omitted to reduce acquisition time. SNR (for b0 images) and CNR (for b1000 images) were calculated with PreQal pipeline[3]. The optimal configuration was further tested with REadout Segmentation Of Long Variable Echo-trains (RESOLVE)[4] DTI to assess reduced artifacts and improved resolution.