Strategies to mitigate gantry angle specific inhomogeneities in a low-field MR-linac.

Document Type

Conference Proceeding

Publication Date


Publication Title

Med Phys


Purpose: Magnetic field inhomogeneities may adversely impact image uniformity and contribute to geometric distortions in MR images. The MR-linac, with 4,000s lbs of steel rotating around a split-field magnet bore, presents unique challenges for maintaining field homogeneity. This work evaluates strategies to mitigate gantry angle specific inhomogeneities, measured using two approaches, to ensure high fidelity imaging. Methods: A 24 cm homogeneous sphere phantom was oriented at magnet isocenter an imaged using the integrated body coil. MFH was evaluated using spectral peak (SP) analysis to yield the spectral full-width at half maximum (FWHM, in Hz) and converted to ppm. To provide a mapping of local inhomogeneity, phase mapping was implemented using dual-echo gradient-recalled echo images acquired with in-phase echo times (20.6/40.3 ms). Phase difference maps were reconstructed after complex division of the complex data and converted to distortion maps using image acquisition parameters. Several strategies were considered for magnet tuning (to yield Gx/Gy/Gz gradient offsets) before being evaluated at 12 gantry positions: gantry angle 0 (G0), gantry 330 (G330, theoretical optimal based on field mapping), gantry-specific offsets (GSO), and mean gradient offsets (MEAN, averaged Gx/Gy/Gz from GSO). To fully characterize the benefits of GSO, SP analysis was conducted every 5 degrees and corresponding phase difference maps were acquired every 30 degrees. Results: MFH depended strongly on gantry angle. Shimming at G0 and G330 yielded unacceptable results (maximum FWHM = 13-20 ppm). MEAN resulted in FWHM of 4.0 ± 1.5 ppm, range: 1.8-7.4. GSO yielded the best field homogeneity, yielding the lowest FWHM across ∼70 gantry angles (1.6 ± 0.4 ppm, range: 1.1-2.3) and >2 ppm at G = 50, 130, 145, 160, and 220-260. Phase mapping highlighted improved inhomogeneity with GSO. Conclusion: Results suggest gantry-specific offsets are optimal for high fidelity MRgRT. Future work will include determining the impact on human images and developing correction schemes.





First Page


This document is currently not available here.