SU-F-J-170: Comparison of Residual Distortions for MRI Sequences On a 3T MRI Unit Using a 3D Rectilinear Grid Phantom.

PURPOSE: Gamma Knife (GK) planning is mostly based upon MRI. Due to the high accuracy required for SRS treatments, geometric distortions in these images should be minimal. MRI Manufacturer provided first order geometric distortions correction algorithms may not reduce distortions to a level that may not be acceptable for treatments requiring high positional accuracy. The purpose of this work is to evaluate the residual, radiosurgery-relevant geometric distortions on VIBE, MPRAGE, GRE and CISS sequences.

METHODS: The 3D rectilinear phantom, GRID3D, was scanned in the Leksell Frame-G with different MRI axial sequences: MPRAGE, VIBE, GRE and CISS, on a Siemens SKYRA 3T MR unit. Siemens 3D distortion correction was applied to all sequences. The residual distortion map of the phantom's 2002 vertices was evaluated by comparing the expected positions of the vertices, with the actual measured locations using Modus' image distortion analysis software (QUASARTM). Distortions were specified as X, Y, Z, and radial deviations. As an independent verification, the MRI sequences were co-registered with CT images of the phantom in GammaPlan.

RESULTS: All sequences showed mean radial residual distortions greater than 1 mm. Maximum radial distortion ranged between 3 and 8 mm. Z axis distortions were the highest for GRE and CISS sequences, and lowest for VIBE. Y distortions were lowest for the VIBE and MPRAGE sequences. X distortion was lowest for GRE and CISS sequences. Almost equal X and Y distortions were measured for CISS. Larger distortions were mainly observed in peripheral regions of the phantom for all sequences. CT-MR co-registration in GammaPlan confirmed these results. The residual distortions were deemed unacceptable for MR based GK treatments.

CONCLUSION: The in-phantom analysis of MR sequences using the Modus phantom allows for effective quantitative evaluation of residual distortion in MR images. MR-CT co-registration facilitates qualitative assessment of MRI distortions.