The idea behind field-map undistortion is that a measured map of the magnetic field actually achieved while the subject was in the scanner is used to correct for distortions field inhomogeneities cause in EPIs.
So, before you can use the tool you need to acquire magnetic field maps.
The basic principle is to measure the change in phase of the MRI signal for two different echo times. The simplest way to do this is to acquire two different images with all parameters identical apart from the TE. Alternatively, you may also wish to use a multi-echo sequence.
This tool requires field maps to be acquired using an non-EPI sequence that does not contain distortions. We think this preferable for two reasons:
(1) Undistortion becomes a 'pull' rather than a 'push' problem: instead of the field map telling you for each point in the distorted space where each voxel should move to, it tells you for each point in the undistorted space where the data have gone.
(2) If you use an EPI sequence you will have least data in some of the regions where you most need it.
The tool requires both phase and magnitude from the field map.
On the Siemens Allegra, you can use the gre_fieldmapping sequence, although for the current version that is shipped, you will need to modify the sequence so that it allows both phase and magnitude to be reconstructed. This is a reasonably simple change.
On a Bruker Medspec scanner, we have used the 3D SPGR sequence with a matrix size of 256x64x64 covering a field of view of 25.6mm cubed. We have used echo times of 3.27 and 10.104 ms.
If you can acquire, reconstruct the phase and magnitude data, and convert them to Analyze data, it should be reasonably simple to modify the tool to accept it. We can offer limited assistance with this final stage.