Click here to

Session: Multi-Disciplinary General ePoster Viewing [Return to Session]

Cross-Validation of MRI Distortion Assessment Using Fiducial and Harmonic Analysis Phantoms

P Yan1*, C Velten1,2, Y Alvarado1, J Anias1, Q Peng1,2, W Tome1,2, (1) Montefiore Medical Center, Bronx, NY, (2) Albert Einstein College of Medicine, Bronx, NY


PO-GePV-M-229 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: Cross-validation of dedicated phantoms for geometric distortion measurement on 1.5T MR scanners.

Methods: Geometric distortion was measured using two MRI phantoms, Magphan RT 820 (The Phantom Laboratory, Salem, NY) and QUASAR MRID3D (Modus QA, London, Ontario), on two 1.5T MRI systems, a Siemens Aera MR simulator and a Philips Achieva.Both phantoms were set up as close to the scanners’ isocenters as possible. For the MRID3D the Philips’ bore size limited vertical setup to be within 5.5 cm. Only the scanner’s internal coil was used for scanning using 3D T1 VIBE (Siemens) and Fast-Field-Echo (Philips) sequences with 1.1mm isotropic voxel size, TR<15ms, TE<10ms, 10° flip-angle, distortion correction enabled, no data interpolation or parallel acquisition techniques. Readout bandwidths of both sequences were 191Hz/px (Philips) and 450Hz/px (Siemens). Quantitative geometric distortion evaluation was performed using vendor-provided software, MRID3D using harmonic analysis and Smári using fiducial-based analysis.

Results: Average absolute distortion components were less than 0.55 mm for both scanners and phantoms, except for y-components measured with MRID3D which ranged from 1.14-1.60mm. Excluding y-components for the MRID3D, the proportion of points with components below a given distortion level (d) within a radial distance (r) from isocenter were for both scanners and phantoms ≥95% for d=1mm/r=10cm, ≥87% and ≥69% for d=1mm/r=20cm for the Magphan and MRID3D, respectively, and 100% for d=2mm/r=20cm.

Conclusion: Distortion levels were found to be consistent between different MRI systems and phantoms, except for a systematic y-offset observed with the MRID3D, that persisted between different MRI systems. In collaboration with the vendor this was found to be due to a manufacturing defect causing the central alignment cuboid to be misaligned with the rest of the phantom. As a result, independent validation of distortion measurements is necessary to ensure the accuracy of the measurement results especially for radiotherapy applications.


Phantoms, MRI, Distortion


IM/TH- MRI in Radiation Therapy: Spatial Distortion/Fidelity

Contact Email