S2093
Physics - Image acquisition and processing
ESTRO 2026
Digital Poster 4470
Evaluation of geometric integrity around hip arthroplasty in prostate cancer radiotherapy simulation using a 0.55T MR Scanner Bertrand Pouymayou, Alessandro Mencarelli, Sebastian M. Christ, Matthias Guckenberger, Lotte Wilke, Stephanie Tanadini-Lang Department of Radiation Oncology, University Hospital Zürich, Zürich, Switzerland Purpose/Objective: Magnetic resonance imaging (MRI) is the preferred modality for delineating prostatic lesions. However, metallic implants from total hip arthroplasty compromise spatial fidelity due to susceptibility artifacts. While such distortions are reduced at low magnetic field strengths, few quantitative data exist on their extend around implants and their correlation in- vivo to B0 field mapping [1]. This study investigates the reliability of gradient echo (GRE) based field mapping to characterize geometric integrity on a low-field MR scanner in the presence of hip arthroplasty implants. Material/Methods: We evaluated the geometric integrity of three sequences: a standard T2 TSE (turbo spin echo, bandwidth (BW)=80Hz/pixel, native resolution 1.2x1.2x3mm3), an optimized T2 TSE for artefacts mitigation (BW=240Hz/pixel , native resolution 1.2x1.2x3mm3) and a T1 vibe Dixon (commonly used for synthetic CT generation, BW=416Hz/pixel, native resolution 2x2x3mm3) acquired on a 0.55T MR scanner (Siemens, Free.Max), using a medium surface coil combined with the spinal coil. Spatial integrity was assessed in the target sequences by imaging a 3D printed grid defining 195 markers (in-plane spacing: 10mm) in the transverse plane. The design allowed placement of metallic implants on both sides, simulating bilateral hip arthroplasties (Figure 1). Five implants with varying magnetic susceptibility from titanium to stainless-steel (Table 1) were tested. Displacements from the reference geometry were measured with a template matching method implemented in Matlab. In each situation, two GRE field mapping sequences (native resolution 2.4x2.4x3mm3) were acquired with 2 (TE=10, 22.94ms) and 3 echoes (TE=6.72, 11.94, 22.46ms), respectively. We used the ROMEO software [2] for phase unwrapping. Distortion maps were obtained pixelwise by dividing the measured B0 (Hz) by the bandwidth and multiplied by the pixel size. For each sequence, we defined a calibrated field map by applying the affine transform minimizing the Euclidian distance between the detected marker positions and the field maps over all implant configurations.
Results: For configurations excluding stainless-steel implants, the mean difference between the detected markers and the calibrated field maps was 0.15(±0.04)mm and 0.15(±0.05)mm for the two- and three-echo methods, respectively (Table 1). Higher readout bandwidth systematically improved agreement between methods and present maximal distortion (95th percentile) below 1mm for non- ferrous alloys.
Made with FlippingBook - Share PDF online