Print

Print
Thankyou, Matt, for your prompt reply!
Just a couple more questions regarding your reply.

1) Eddy_Correct: So, when I begin the fdt processing stream, I first run eddy_correct.  eddy_correct works on a 4d file with the nodif image simply being the first B0 volume.
That B0 volume is not BET extracted, nor is anything else in the 4D image BET extracted.
Is this correct and as it should be?

Next I use avwsplit to get the separate volumes in the 4D file, and I save the 1st volume (my first B0, and the one I registered to in step 1)  as no_dif.nii.gz

2) Bet2: Next I run bet2 on nodif.nii.gz to get the no_dif_brain file and the corresponding nodif_brain_mask file. 

3) DTIFIT: Then I run dtifit using my BET extracted brain mask, no_dif_brain_mask

4) Bedpost: Now I run bedpost using the no_dif_brain and the no_dif_brain_mask, both of which are bet2 extracted already.

5) Registration: Now I run registration, I need to run Bet on my structural image, BUT I already have a Bet extracted no_dif_brain that I created in step 2 and used in step 4.  Is that no_dif_brain file the one to use here in registration, or do I, for some reason, need to extract ANOTHER copy of the no_dif_brain from my original 4D data file?

Thankyou so much for helping me get this straight...

-Dianne

On 5/23/07, Matt <[log in to unmask]> wrote:

Dianne,

 

You want to extract the nodif brain before running FDT, BEDPOST, or any registrations.  Additionally you need to make a brain only mask nodif_brain_mask before running either FDR or BEDPOST, if you don't it will try to calculate outside of the brain which will just waste time and produce junk outside.  You can use the commandline "bet nodif nodif_brain –m –f .2" to make that mask.  You can adjust the last number, .2 to get better extractions (lower is more conservative, higher more liberal).  Also registrations will be most successful on extracted brains, as they rely on the edge of the brain to make the alignments.  

 

Peace,

 

Matt.

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Dianne Patterson
Sent: Wednesday, May 23, 2007 1:21 PM
To: [log in to unmask]
Subject: [FSL] Just trying to get this straight

 

For the fdt pipeline:

1) When I run bedpost, my nodif image should NOT be brain extracted (i.e., it should still have the skull on?)
(what, if anything, will go wrong if no_dif IS brain extracted?)

2) However, when I run registration, my nodif image and my struct should be brain extracted versions.

Thanks,

Dianne

--
Dianne Patterson, Ph.D.
[log in to unmask]
ERP Lab
University of Arizona
621-3256 (Office)




--
Dianne Patterson, Ph.D.
[log in to unmask]
ERP Lab
University of Arizona
621-3256 (Office)