Quantitative Evaluation of MRI Geometric Distortion from 1.5T to 7T
Abstract
Purpose
This study is to quantify geometric distortion magnitude and evaluate vendor-provided distortion correction performance across four Siemens and GE MRI scanners operating at 1.5T, 3T, and 7T.
Methods
A grid distortion measurement phantom (Modus Medical Devices) was scanned on a Siemens Sola 1.5T, Siemens Skyra 3T, GE Signa Premier 3T, and Siemens Terra.X 7T. Scans used a 3D T1 MPRAGE sequence at 200 Hz/pixel bandwidth, 2520 ms TR, 2.19-3.4 ms TE, 4º flip angle, 1500 ms TI, 256x256 matrix, 1x1x1 mm voxel size, and sagittal view. Three sets of images were taken for each scanner: no distortion corrections, 2D distortion corrections, and 3D distortion corrections. Distortion measurements were calculated using QUASAR GRID3D, a software built to analyze the phantom. Values from the software were exported to Matlab. Comparisons were made with the average absolute distortion using 27 points distributed between the middle and edge slices of the phantom, and the 95th percentile (P95) of distortion values up to 100 mm from isocenter.
Results
Uncorrected average absolute distortion was 1.13 mm (Siemens 1.5T), 0.93 mm (Siemens 3T), 0.97 mm (GE 3T), and 1.49 mm (Siemens 7T). With 2D correction, averages were 0.54 mm, 0.78 mm, 0.97 mm, and 0.78 mm, respectively, while 3D correction yielded 0.46 mm, 0.84 mm, 0.98 mm, and 0.74 mm. Uncorrected P95 distortion values were 1.47 mm, 1.37 mm, 1.56 mm, and 2.37 mm. With 2D correction, values were 1.53 mm, 1.22 mm, 1.59 mm, 1.92 mm. 3D correction reduced values to 0.68 mm, 1.08 mm, 1.56 mm, and 1.23 mm.
Conclusion
Distortion increases with field strength and with distance from isocenter. 3D distortion correction techniques show the greatest improvement to measured values across all scanners. Siemens correction techniques make a larger impact than their GE counterparts. Results show potential for system calibrations for clinical practice.