Introduction
This is a work-in-progress. Please check back for updates, and check software carefully before using!Gridding is a method of interpolating data from an arbitrary 2D sampling pattern to a uniform grid. In MRI, this allows rapid image reconstruction. There are numerous publications on MR gridding reconstruction - I'll try to add these soon!
Included on this site are some Matlab gridding functions that will help you get started with gridding reconstructions. I have avoided making generalizations about what method is best(!)
Functions Included
The tar file includes numerous .m scripts, and C code that can be compiled to a .mex function.Useful Matlab Functions:
- gridkb.m -- Matlab gridding function that does most of what you need. (Requires other functions!)
- plotgridkb.m -- Nice utility to plot the grid samples, and the convolution of samples with the kernel to check your parameters.
- plotgridkbkernel.m -- Utility to plot the convolution kernel used by gridkb.m finely.
- plotlutkernel.m -- Similar Utility to plot the kernel passed to gridlut.m kernel finely.
- calcdcflut.m -- Calculate the density compensation factors using a lookup-table. (We still need a "calcdcfkb.m" ).
Support Functions:
- gridlut.m -- Gridding, using an arbitrary kernel-lookup. (Calls gridlut_mex, which is a compiled MEX function whose extension depends on the operating system used.)
- kb.m -- Calculate the Kaiser-Bessel function used in gridding kernel calculation. (Called by calckbkernel.m)
- calckbkernel.m -- Calculate the Kaiser-Bessel kernel.
C and MEX functions:
- gridlut_mex.c -- MEX interface to C function for gridding.
- calcdcflut_mex.c -- MEX interface to C function for calculating the density compensation factors for gridding.
- gridroutines.c -- General C functions for gridding.
This is a sample illustrating how gridding convolves arbitrary sample points (in this case a spiral) with a kernel. The convolved data are then sampled onto a uniform grid (shown by small dots) so that they can be Fourier transformed using the FFT.
