System and method for image warp correction for magnetic resonance imaging
A method of correcting warping of an acquired image in an MRI system, caused by non-linearities in gradient field profiles of gradient coils is set forth, comprising a) constructing a computer model representing conducting pathways for each gradient coil in said MRI system; b) calculating a predicted magnetic field at each point in space for each said gradient coil in said model; c) measuring actual magnetic field at each point in space for each said gradient coil in said MRI system; d) verifying accuracy of said model by comparing said predicted magnetic field to said actual magnetic field at each said point in space and in the event said model is not accurate then repeating a)-d), and in the event said model is accurate then; constructing a distortion map for mapping coordinates in real space to coordinates in warped space of said acquired image based on deviations of said predicted magnetic field from linearity; and unwarping said warping of the acquired image using said distortion map.
1. A method of correcting warping of an acquired image in an MRI system, caused by non-linearities in gradient field profiles of gradient coils, comprising:
constructing a distortion map for mapping each location x, y, z in real space to a location x′,y′,z′ in warped space based on deviations from linearity of a magnetic field produced by the gradient coils;
wherein image values at x′, y′, z′ are computed directly from k-space data using inverse gridding from uniform to non-uniform sample locations; and wherein
(i) k-space data is transformed to uniformly spaced samples in warped space,
(ii) image values are computed at the x′, y′, z′ warped space locations using convolution interpolation, and
(iii) the computed image values are used to form a real space image.
2. The method of claim 1 , wherein said mapping maps uniformly spaced samples in real space to non-uniformly spaced samples in warped space.
3. The method of claim 1 wherein said inverse gridding is effected by performing an apodization correction in k-space followed by an inverse Fourier transform.
4. The method of claim 1 further comprising zero padding k-space to decrease spacing between uniformly spaced samples in warped space.
5. The method of claim 4 wherein the spacing is decreased by a factor of 1-2.
6. The method of claim 1 wherein locations of the acquired image in warped space are more densely spaced than the uniformly spaced samples in warped space.
7. The method of claim 1 , where the convolution interpolation uses a weighting value applied to each location x′, y′, z′ to account for non-uniform voxel size in real space.