2008/4/16, Mark Jenkinson <[log in to unmask]>:

Dear HT,

You don't need to have the images exactly matched, but there are
often issues with the alignment when using different FOVs and resolutions
due to the different conventions that people use (for origin etc.) Therefore
it is probably best to use the FEAT pre-processing pipeline for this, as it
does a lot of registration steps in order to correct for these things.
Alternatively, just resample your fieldmap using flirt with the epi_magnitude
as a reference image and an identity transformation. This should give you
some idea of whether they are well matched or not. Note that any constant
B0 fields introduced by the shims will shift the distorted images with respect
to the non-distorted ones and this may be hard to account for outside of the
FEAT-style registration pipeline.

As for not seeing the corrections, this may depend on your units. Are you
calculating your fieldmap in radians per second? That is, if you TE shift is
2 ms (say) then you need to take the phase difference and divide it by 0.02
seconds. This normally gives values in the range of 100's to 1000's in the
rad/s fieldmap for a human brain. If you can see the distortions in your
phantom EPI, I'd expect a similar range, assuming that your EPI has fairly
typical (for human brain) settings. Also, make sure your dwell time is
specified in seconds.

I hope that's enough to sort out your problem.
If not, let us know and send more details.

All the best,
Mark

On 16 Apr 2008, at 01:18, HT wrote:

Hi, I am trying to use the fugue function to correct off resonance effect of
an EPI image using the following line:

fugue -i epi_magnitude --dwell=dwelltime --loadfmap=fieldmap -u result

I have several questions regarding this line:

1. Does the epi_magnitude image need to have the same matrix size as the
field map? That is, if my fieldmap is 256 x 256 and my epi_magnitude is 64 x
64, does fugue downsample my fieldmap to 64x64 for me or I need to do that
myself?

2. The fieldmap is computed from two phase images of fGRE(256 x 256) with a
TE shift and unwarped using prelude. The matrix size of the epi image is 64
x 64, both were acquired with miscalibrated xy shim with a water phantom.
The epi image is clearly distorted and however when i run this line, I
didn't see any corrections in the result image. Am I using fugue correctly
as it is intended?

Since the data is acquired with a phantom and there is no motion involved, I
have skipped steps of coregistration and etc.

Any help or suggestions would be great! Much appreciated your help!