Hi,
The TE difference is the difference in echo times between the two different images that are acquired for the fieldmap sequence (and has nothing to do with the EPI - the EPI settings are used in the FEAT GUI, but not in fsl_prepare_fieldmap). Note that each image within the fieldmap sequence is associated with both a magnitude *and* phase image, but the reconstruction that occurs on the console often does some processing and only returns the difference of the two phase images. However, it is incorrect to say that it is the difference in TE between the magnitude and phase images, and each magnitude has an associated phase image that it acquired simultaneously (they are both part of a complex data pair). This is a minor point for your understanding, but it sounds like the value you are using is the correct one (although you have not said what the value actually was).
The orientation is tricky. If the phase encode direction was A>>P and you can see that the y axis (in *voxel* coordinates) is the A-P axis then this means it is either y or -y. However, due to the various possibilities for things to change direction and sign (e.g. in the phase difference) within the console reconstruction, conversion to nifti, and subsequent pre-processing, I have found that it is unreliable to try and figure out the sign from first principles. Instead it is better to try both y and -y in your analyses and then decide which actually works well (the other one will make things worse, not better). Once you've done this for one set of data from your scanner, all other datasets using the same protocol will have the same sign, so you only need to figure this out once for each protocol that you set up.
The Jacobian message is most likely to be generated by the non-linear registration. However, it may not be problematic. Check the registration results carefully (best done with fslview) and see if you can see any problems or not.
All the best,
Mark
On 24 Apr 2014, at 21:44, Marco Solmi <[log in to unmask]> wrote:
> Hi,
>
> I tryed to use fieldmap unwarping in the preprocessing, for fMRI data.
> Data come from SIEMENS MAGNETOM Skyra syngo MR D11
> For the EPI data I used echospacng 0.55 and TE 30ms
> while for the fieldmap I used the difference between TE of the magnitude and of the phase field images, to create the rads image with the fsl_prepare_fieldmap SIEMENS command.
> As orientation I used -y (actually that scan says that the pahse encoding direction is A>>P).
>
> So I would ask 2 questions:
> Are the orientation and the way of computing the TE difference correct ?
>
> What could the following message I got in log_report be due to?
> "Warning, Jacobian not within prescribed range. Prescription is 0.01 -- 100 and obtained range is -0.0042974 -- 7.88707"
>
> Thanks a lot for your time.
>
> Best
|