I haven't done much with trackvis, but can't you transform the result into a
binary mask and then move it to standard space with applywarp (use nearest
neighbor interpolation)? I did this with the predecessor to trackvis (DTI
TaskCard).
Peace,
Matt.
-----Original Message-----
From: lamberton franck [mailto:[log in to unmask]]
Sent: Monday, June 28, 2010 10:46 AM
To: FSL - FMRIB's Software Library
Cc: Matt Glasser
Subject: Re: [FSL] non linear transformation of deterministic tractography
Thank for the response.
Le 28/06/10 15:04, Matt Glasser a écrit :
> What package are you planning to use for deterministic tractography?
At the moment, I used the command line 'dti_tracker' from Diffusion
Toolkit. Results are stored in a .trk file which is clearly explained on
the Website of the software
(http://www.trackvis.org/docs/?subsect=fileformat)
> Why
> can't you just transform the results into standard space?
>
Maybe I'm confused and my question is very simple.
"Deterministic tracts" are constituted by a series of points defined by
their 3 coordinates x/y/z. Linearly transforming these series of points
is very easy : one've just to apply the transformation matrix (4x4
matrix computed with FLIRT) to each point (column vector with the 3
coordinates x/y/z followed by 1).
Conveniently, Diffusion Toolkit contains a command line
'track_transform' which allows to do that directly on .trk file.
However, I don't see how I can apply a non linear tranformation such as
computed by FNIRT on a series of points defined by their 3 coordinates.
Using FNIRT, I managed to non linearly transform a volume (ex: T1,
probabilistic tracts, ...) but not a series of distintcs points defined
by their coordinates x/y/z as the output f 'dti_tracker'.
Sincerely
Franck
> Peace,
>
> Matt.
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
> Of lamberton franck
> Sent: Monday, June 28, 2010 2:47 AM
> To: [log in to unmask]
> Subject: [FSL] non linear transformation of deterministic tractography
>
> Hi
>
> In order to analyze micro- and macrostructure of white mater, I want to
> apply and compare a deterministic and a probabilistic tractography
> method on a set of DW-MRI.
>
> Excepted the deterministic tractography, all the processing was already
> done with FSL package : diffusion analysis was performed in the
> diffusion space and results such as diffusion parameters (FA, L1, ...)
> and probabilistic tracts were transformed in the standard space using a
> non-linear transformation computed with FNIRT.
>
> My question is related to the transformation of "deterministic" tracts
> from the diffusion to the standard space.
> A method could consist in using "vecreg" in order to transform V1
> volumes into the standard space and then applying a deterministic
> tractography method. However, this implies an interpolation step that I
> don't want.
> Does anybody have any experience on applying non linear transformation
> computed with FNIRT to a "deterministic" tracts ?
>
> Thanks in advance for any help.
>
> Franck
>
>
|