Hi Matt,

I am having a hard time understanding some of the convertwarp flags regarding what we have discussed last week. I want to register my second session to the first and then apply the combined warp generated from the first session that to that image to get it into the template image, as you suggested. I am not sure if I am using the right flag designations for each of the matrices when using convertwarp, as there seem to be multiple ways to do this. Here's what I've got so far:

flirt -ref subj1_sess1_FA.nii.gz -in subj1_sess2_FA.nii.gz -omat subj1_sess2_aff.mat
convertwarp --ref=subj1_sess1_FA.nii.gz --warp1=subj1_sess1_trans.mat.nii.gz --postmat=subj1_sess2_aff.mat --out=subj1_comprehensive_warp

subj1_sess1_trans.mat.nii.gz is what I got from the nonlinear registration of the first session to the standardized space. 

Are warp1 and postmat the right flags for their respective matricies? 

Thanks for your time.

Namik 



On Apr 30, 2010, at 5:18 PM, Matt Glasser wrote:

For my first session, I do the 6 dof, 12 dof, and nonlinear registrations all separately, however I then combine them all into a signal warpfield using convertwarp to simplify things.  I am suggesting that you use 6 dof to register your second session to the first and then apply the combined warp you generated with the first session data to that image to get it into template space.  I found that for a dataset where the second session patients had a surgical lesion, that significant nonlinear distortions were introduced by debulking the tissue, and that a 6 dof registration and a nonlinear registration were both necessary (there were not, however, and global skews or scales so I left out the 12 dof part).  

 

Peace,


Matt.

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Namik Kirlic
Sent: Friday, April 30, 2010 4:10 PM
To: [log in to unmask]
Subject: Re: [FSL] FA registration query

 

Thank you, Matt.

 

So, does this mean that when we use flirt to register the first session, we should use a 6-parameter rigid body registration, and then apply the first session's affine registration onto second session?

 

For example: 

 

 

Subject-1 session-1

 

flirt -ref FMRIB58_FA_1mm -in subj1_sess1_FA.nii.gz -omat subj1_sess1_aff.mat -dof 6

fnirt --in=subj1_sess1_FA.nii.gz --aff=subj1_sess1_aff.mat --cout=subj1_sess1_trans.mat --config=FA_2_FMRIB58_1mm.cnf

applywarp --ref=FMRIB58_FA_1mm --in=subj1_sess1_FA.nii.gz --warp=subj1_sess1_trans.mat.nii.gz --out=subj1_sess1_warp_FA

 

 

Subject-1 session-2

 

fnirt --in=subj1_sess2_FA.nii.gz --aff=subj1_sess1_aff.mat --cout=subj1_sess2_trans.mat --config=FA_2_FMRIB58_1mm.cnf

applywarp --ref=FMRIB58_FA_1mm --in=subj1_sess2_FA.nii.gz --warp=subj1_sess2_trans.mat.nii.gz --out=subj1_sess2_warp_FA

 


Thanks. 

 

 

 

On Apr 29, 2010, at 11:37 PM, Matt Glasser wrote:



I agree it makes sense to use the same nonlinear transform (in fact I have done something similar for a particular dataset).  It should be sufficient to use 6-parameter rigid body registration for two sessions of the same subject, unless something significant has changed (e.g. a surgical lesion, which can itself require a nonlinear transform to the previous dataset, which FNIRT happens to be quite good at as it turns out).  Are you using one of the FSL default analysis pipelines like TBSS?

 

Peace,


Matt.

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Namik Kirlic
Sent: Thursday, April 29, 2010 3:42 PM
To: [log in to unmask]
Subject: [FSL] FA registration query

 

Hello,

We are registering FA images for subjects who have been scanned across
multiple imaging sessions. Given that we hypothesize that FA (but not the macrostructure of the brain) is going to change across sessions, this could potentially affect the non-linear (i.e. fnirt) component of registration. We were wondering if there is any way to separately register the images from the two sessions with flirt, but to then apply the same non-linear transform (i.e. from one of the sessions) to both sets of images. We are asking both whether this makes theoretical sense, and practically speaking, how it could be done, given that the fnirt stage is done in conjunction with flirt and we are not certain if the non-linear component can be separated out as its own transform to be applied.

I should also mention that part of the reason we would like to do this is that we are also moving some ROI segmentations drawn in standard space back to native space (using an inverse transform) for probtrack analysis. When we load the segmentations onto the FA images in native space, they do not look the same across sessions. They look similar, but definitely not identical in shape or even in size.

 

 
 
The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.