 |
 |
Hi, I would align your reference epi (used with reffile) to the structural using epi_reg only once. Then you can apply the transformation (a warp-field or a matrix, depending on if you have or have not used fieldmaps) to get all the functional images (that are aligned with the reffile) into the structural space. You can combine the motion correction matrices with the transformation from epi_reg using applywarp. There is no separate output option in epi_reg for the transformations - but they are always saved. Look for .mat or warp files that use your output basename. All the best, Mark On 27 Nov 2014, at 20:19, Annelinde Vandenbroucke <[log in to unmask]<mailto:[log in to unmask]>> wrote: OK, one more question. If I do motion correction to one reffile, and thus epis are aligned, do you think it's better to use epi_reg on each run separately to register the functionals to the structural, or should I make one transformation matrix for the 1st run using flirt, and then apply this matrix to all other runs using flirt again? It seems epi_reg does not have an option to save a transformation matrix (because it uses segmentation to coregister?) that I can then use on all functionals. Thx! Annelinde 2014-11-27 7:36 GMT-08:00 Mark Jenkinson <[log in to unmask]<mailto:[log in to unmask]>>: Hi, There are a couple of different options here: (i) use the -reffile option in mcflirt if the runs aren’t too different in positioning and this works OK (ii) perform motion correction separately and then use flirt to register the individual reference volumes together and then apply the motion correction matrices together with these flirt matrices to do a single resampling from the original images to the final corrected version. All the best, Mark On 26 Nov 2014, at 18:32, Annelinde Vandenbroucke <[log in to unmask]<mailto:[log in to unmask]>> wrote: Well, the main thing is that I would like to do motion correction so that all the runs after run 1 are aligned to the first run, such that they are exactly in the same space. I'm afraid that if you register them all to the anatomical separately, small differences will occur such that voxels are not exactly in the same space. I'm going to do reconstruction using voxel patterns, so it is very important that all functionals are in exactly the same space. 2014-11-26 9:51 GMT-08:00 Bryson Dietz <[log in to unmask]<mailto:[log in to unmask]>>: Why not just register your functionals to anatomical space (epi-reg function) (therefore they are registered to one another) then re-sample back to its original space? Or am I missing the point of what your trying to achieve? On Wed, Nov 26, 2014 at 12:18 PM, Cinly Ooi <[log in to unmask]<mailto:[log in to unmask]>> wrote: Presuming you are talking about 1st level, what I do is to create a seed fsf file using one subject then write a script to copy the seed fsf file over and change it for other subjects. The alternative, should you incline to do, is to look at reproducing report_log.html using scripts. Best Regards, Cinly ***** Don't bother with footer please. I don't read them and will not be bounded by them. It cannot be enforced legally anyway. If it can, then remember this: This footer always triumph yours. On 26 November 2014 at 16:45, Annelinde Vandenbroucke <[log in to unmask]<mailto:[log in to unmask]>> wrote: Hi Andreas, Thank you for that! I was actually wondering whether I can do it in script in stead of Feat. Cheers, Annelinde 2014-11-26 0:12 GMT-08:00 Andreas Bartsch <[log in to unmask]<mailto:[log in to unmask]>>: Hi, you can use the alternative reference image in the preprocessing tab / motion correction within Feat. Due to the fatsat the difference in registering unBETed vs. BETed EPIs to the structural is often small. Cheers, Andreas Am 26.11.2014 um 00:01 schrieb Annelinde Vandenbroucke <[log in to unmask]<mailto:[log in to unmask]>>: > Hi, > > I have an experiment with several functional scans that I would like to align to each other, and then align them to the structural scan (stay in native space though). Preferably, I would align run 2-n to run 1, and then have the same transformation to coregister to the structural scan. I know that in SPM and Brainvoyager, it is possible to do motion correction on run 2-n aligned to the first volume of run 1, but I was wondering how to do that in FSL. > > Also, should I first use a betted functional to coregister to my structural, and then use this matrix to coregister the unbetted functional? Or does EPI-align work well with the functional scans that are not betted? I can then mask the functional scan after coregistration > > Thanks! -- Annelinde R. E. Vandenbroucke, PhD University of California, Berkeley || Helen Wills Neuroscience Institute 10 Giannini Hall Berkeley, CA 94720 -- Annelinde R. E. Vandenbroucke, PhD University of California, Berkeley || Helen Wills Neuroscience Institute 10 Giannini Hall Berkeley, CA 94720 -- Annelinde R. E. Vandenbroucke, PhD University of California, Berkeley || Helen Wills Neuroscience Institute 10 Giannini Hall Berkeley, CA 94720