Gradient Removal in 3D MR Images
Retrospective Model-Based Reduction of Gradients in MR Images
(follow link for PDF version)
Technical Paper—1995
William T. Katz, Ph.D., M.D.
Neal F. Kassell, M.D.
James R. Brookeman, Ph.D.
Operations using image intensities often assume substances have spatially invariant intensities. This assumption may be incorrect in magnetic resonance imaging due to low spatial frequency intensity variations introduced by the radio frequency coil. This paper describes a method for indirect calculation of the coil sensitivity profile which uses the imaged brain as a qualitative phantom. The algorithm estimates ideal image intensities for likely brain voxels in a subsampled fashion, calculates an averaged sensitivity profile at each point in the subsampled grid, and then fits an order two three-dimensional polynomial to the averaged sensitivities. Since a voxel-based anatomic model is used to select the brain data points, user interaction is reduced to the initial registration of the image to Talairach space.
Key Words: Retrospective gradient removal, Model-based method, Three-Dimensional imaging.
Note: Images for this paper are currently not available.
