Dear Bastian,
> Dear Jesper,
>
> Thanks a lot for your reply, it already improved the outcome! In fact the unwarp direction I used was not correct. Nevertheless, it produces now a new warping-problem: The unwarping produces a "hole" consisting of negative values in the orbito-frontal part of the diffusion data. I uploaded again an exemplary image together with the registered field map and magnitude image at those coordinates. Do you have any idea what might be the problem here?
>
this is something that can happen when the distortions are sufficiently large that parts of the brain “collapses onto itself”. Specifically, if the spatial rate of change of the off-resonance field is very large combined with a slow traversal of k-space (e.g. high resolution, no IPAT etc) you can end up with a situation where signal that should be more anterior than some other signal ends up being more posterior than that other signal in the reconstructed image. This is what we refer to as the “Jacobian becoming negative”.
The model that is used for the resampling is an approximation that breaks down when the Jacobian becomes sufficiently small or becomes negative, and you end up with the negative intensities that you see.
The reason that you don’t see this with topup-generated fieldmaps is that, because of the breakdown of the Jacobian model, you will never estimate a field that causes such a collapse. Which is not to say that such a collapse hasn’t occurred. It really is a problem that originates from the acquisition.
My advice is to ignore those areas in the subsequent analysis.
Jesper
> My routine (based on Matt's code mentioned before and your suggestions) looks the following:
>
> 1) Betting the magnitude image
> 2) fsl_prepare_fieldmap
> 3) Betting the field map
> 4) Smoothing the field map
> 5) Eddy-current correction
> 6) Warping the warped magnitude image using the field map to match the distortion in the diffusion data
> 7) Registering the magnitude and field map image to the first b0
> 8) Unwarping the data with the registered field map
>
> Might there be a problem in my routine, that produces this "hole" of negative values? One also has to mention, that the brains described by the DTI data contain large hemispheric lesions, could this affect the unwarping?
>
> Again, thanks already in advance for your help!
>
> Best,
> Bastian
>
> P.S.: Sorry if this is a double-post now, I sent the message yesterday already but it never appeared on the list.
>
> <warped_data.png><magnitude_2_data.png><fm_2_data.png>
|