 |
 |
No, I'm almost positive that was something after fnirt was applied to a brain with lots of neck and other non-brain included. Joelle would have to answer that, since I couldn't reproduce that image, exactly. The original T1 has a lot of neck with a little shoulder, which is why brain extraction was difficult. I got something like this with a bad brain extraction (after fnirt) [image: Inline image 1] With a better brain extraction I got this (again, fnirt) [image: Inline image 2] Cheers, Jeanette On Fri, Jun 5, 2015 at 3:45 PM, Andreas Bartsch <[log in to unmask]> wrote: > Hi, > > but the screenshot did not look like any affine transformation. > Was this really flirt? > Cheers, > Andreas > > Von: Jeanette Mumford <[log in to unmask]> > Antworten an: FSL - FMRIB's Software Library <[log in to unmask]> > Datum: Freitag, 5. Juni 2015 16:14 > An: <[log in to unmask]> > Betreff: Re: [FSL] Registration of 4D time series to MNI standard > > Just to update the list, so far the culprit was bad skull stripping, but > adding the -R flag to bet improved skull stripping and then the resulting > flirt registration. > > Cheers, > Jeanette > > On Thu, Jun 4, 2015 at 4:04 PM, Jeanette Mumford < > [log in to unmask]> wrote: > >> Hi Joelle, >> >> This is the flirt output? That's a 12DOF linear transform, so I wouldn't >> think the front of the brain would look like that unless it did in the >> original image. Also, a skull stripped image should be used with flirt. >> >> Off the list (email me directly), can you email me your original T1 >> structural (with skull) and I'll have a look. >> >> Cheers, >> Jeanette >> >> On Thu, Jun 4, 2015 at 3:30 PM, Joelle Zimmermann < >> [log in to unmask]> wrote: >> >>> Hi Jeanette, >>> >>> First of all, thanks for your help. That's very nice of you. Yes, I >>> think the first flirt step (in email above) is already giving the wrong >>> output. I am attaching a screenshot, just of the output - the >>> highres2standard. >>> >>> You can see that it is sort of leaking from the top. >>> >>> I think the orientation of the T1 is correct, it looks like it's >>> centered on the field of view. The brain.nii itself looks fine to me, but >>> I'm no expert. >>> >>> Do you have any ideas of what to do to correct this first flirt step? >>> I'm really new to this, I've been struggling with this whole registration >>> process - didnt realize it was such an art! :-P >>> >>> Thanks, >>> Joelle >>> >>> On Thu, Jun 4, 2015 at 3:24 PM, Jeanette Mumford < >>> [log in to unmask]> wrote: >>> >>>> Hi Joelle, >>>> >>>> I'll start here, since if this step doesn't work the others likely >>>> won't either. I'm unsure why they are shifted, I haven't encountered this >>>> before. Is the orientation of your T1 correct before registration? Is it >>>> basically centered in the field of view? Also, is this a typical adult >>>> brain? >>>> >>>> Cheers, >>>> Jeanette >>>> >>>> On Thu, Jun 4, 2015 at 3:13 AM, Joelle Zimmermann < >>>> [log in to unmask]> wrote: >>>> >>>>> Thanks very much Jeanette. That's very kind of you. I notice that >>>>> after I run the flirt step: >>>>> >>>>> flirt -in bet_highres -ref $FSLDIR/data/standard/MNI152_T1_2mm_brain >>>>> -out highres2standard -omat highres2standard.mat -cost corratio -dof 12 >>>>> -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -interp trilinear >>>>> >>>>> And I overlay the MNI152_T1_2mm_brain standard over the >>>>> highres2standard output, it doesn't give me a good fit. Which I think it >>>>> should. They are a bit shifted from each other. Do you have any suggestions >>>>> of how to fix this? >>>>> >>>>> Thanks, >>>>> Joelle >>>>> >>>>> >>>>> >>>>> >>>>> On Wed, Jun 3, 2015 at 5:36 PM, Jeanette Mumford < >>>>> [log in to unmask]> wrote: >>>>> >>>>>> See my responses within yours below.... >>>>>> >>>>>>> >>>>>>>> fslroi bold_data example_func 124 1 >>>>>>>> >>>>>>>> McFlirt: Motion correction (skip this if you already ran it) >>>>>>>> mcflirt -in bold_data -out bold_data_mcf -mats -plots -reffile >>>>>>>> example_func -spline_final >>>>>>>> >>>>>>> >>>>>>> is the -in bold_data the 4D functional time series? >>>>>>> >>>>>> >>>>>> Yes, it is. >>>>>> >>>>>> >>>>>>> >>>>>>>> Flirt: Register EPI to Subject's structural. >>>>>>>> epi_reg --epi=example_func --t1=your_highres_with_skull >>>>>>>> --t1brain=highres_no_skull --out=example_func2highres >>>>>>>> # This will create many files, but you only need to keep a file >>>>>>>> called example_func2highres.mat. >>>>>>>> >>>>>>>> Flirt/Fnirt combo: Structural to MNI >>>>>>>> flirt -in bet_highres -ref >>>>>>>> $FSLDIR/data/standard/MNI152_T1_2mm_brain -out highres2standard -omat >>>>>>>> highres2standard.mat -cost corratio -dof 12 -searchrx -90 90 -searchry -90 >>>>>>>> 90 -searchrz -90 90 -interp trilinear >>>>>>>> >>>>>>> >>>>>>> The -in bet_highres parameter, is that really the same as the >>>>>>> highres_no_skull from the previous step? bet_highres is just the high res >>>>>>> T1 with the skull removed, right? >>>>>>> >>>>>> >>>>>> Yes, sorry I was inconsistent, bet_highres it is the highres without >>>>>> skull. >>>>>> >>>>>>> >>>>>>>> fnirt --iout=highres2standard_head --in=your_highres_with_skull >>>>>>>> --aff=highres2standard.mat --cout=highres2standard_warp >>>>>>>> --iout=highres2standard --jout=highres2highres_jac --config=T1_2_MNI152_2mm >>>>>>>> --ref=$FSLDIR/data/standard/MNI152_T1_2mm --refmask=standard_mask >>>>>>>> --warpres=10,10,10 >>>>>>>> >>>>>>> >>>>>>> is there a mistake or should there really be two iouts? >>>>>>> (highres2standard_head and highres2standard). >>>>>>> >>>>>> I grabbed this command directly from a FEAT log and it does have two >>>>>> --iout flags. This flag is simply the name of the output image. I'm >>>>>> unsure why it is in there twice, but presumably it will work fine. If you >>>>>> run it this way, you can see if it creates 2 output files or simply uses >>>>>> the last name >>>>>> >>>>>>> >>>>>>> what is the --refmask=standard_mask? I suppose it would be the MNI152_T1_2mm_brain_mask_dil.nii >>>>>>> that is supplied by fsl? >>>>>>> >>>>>> >>>>>> It is $FSLDIR/data/standard/MNI152_T1_2mm_brain_mask_dil >>>>>> >>>>>> Hope that does the trick! If you want you can run a "prestats only" >>>>>> Feat analysis via the GUI and look in the log file to see the commands that >>>>>> were used. For the most part you can simply reuse them. >>>>>> >>>>>> Cheers, >>>>>> Jeanette >>>>>> >>>>> >>>>> >>>> >>> >> >