Geometric accuracy of surface high-dose rate brachytherapy applicator and catheters identified using Dixon volumetric interpolated breath-hold examination magnetic resonance images

Background and purpose: Volumetric Interpolated Breath-hold Examination (VIBE) magnetic resonance imaging (MRI) can depict flap applicators of High-Dose-Rate (HDR) surface brachytherapy. Evaluating criteria for catheter detection and assessing the geometric fidelity of detected applicators and catheters is important for VIBE implementation in HDR brachytherapy planning. Materials and methods: A fixed flat Freiburg Flap (FF) applicator (31 empty catheters, 21 spheres each) was imaged axially in a computed tomography (CT) scanner and 3 T MR simulator using Dixon VIBE. VIBE was repeated for three sagittal and coronal applicator placements. A novel algorithm detected applicator borders and catheters on central slices of in-phase (IP) and opposed-phase (OP) series, their geometric mean (GM), and reference CT scans. Obtained distances were compared against known dimensions. Applicator and catheter edge displacements between IP and OP images were measured. Results: All catheters were automatically identified on all series. Applicator lengths generally lay within 310 ± 1 mm. Catheter lengths were underestimated by up to 2 mm on original images due to reliance on surrogate tunneled beads. Adjacent catheter positions lay on average within 0.05 mm from their nominal 10.0 mm distance. Catheter angles were opposite between IP and OP images affected by chemical shift. GM corrected the relative applicator and catheter rotation. Average chemical shift induced IP-OP displacements were 2.3 ± 0.6 mm. Conclusions: Local signal intensity minima were reliable for empty catheter detection. Applicator shape and dimensions were preserved on individual IP and OP series. Chemical shift induced displacements amounted to 1.2 mm for individual images.