Thanks, Steve. You're right, I meant to write that I'd align to the nodif image. I will try the new fslmaths command and see if it works, but it looks like it should do the trick. I'll try again to add the structures in the JHU white matter atlas and see what happens. When I tried it, it showed the individual structures on the list, but then when I selected it, it brought up a mask with all labeled structures. I tried saving that mask as a file, and the result was the same mask with all structures labeled except that the tract of interest was a different color. Maybe I'm using an older version of FSLview? I'm using FSL 4.0, but it may not be the newest version. Thanks again for all of your help... you're the best!
Kristen
-----Original Message-----
From: FSL - FMRIB's Software Library on behalf of Steve Smith
Sent: Mon 2/25/2008 3:41 AM
To: [log in to unmask]
Subject: Re: [FSL] DTI-based White Matter Atlas
Hi,
On 23 Feb 2008, at 22:11, Lindgren, Kristen, Ann wrote:
> Thanks for the info, Steve. Here's what I've done. Please let me
> know if this sounds correct.
>
> 1) I've preprocessed all of the DTI data through FDT (eddycorrect,
> dtifit, and bedpostx)
>
> 2) Using the FDT Registration GUI, I aligned dti_FA.nii.gz to both a
> high-res anatomical image and the default standard brain.
OK - though note that if you used the FDT GUI it is the b=0 ('nodif')
image that is actually aligned to the hires image.
> 3) Using ApplyXFM_gui, I selected diff2standard.mat as the transform
> matrix, dti_FA.nii.gz as the input volume, and the MNI brain used in
> the FDT registration as the reference volume and made a
> dti_FA_alignstandard.nii.gz file.
>
> 4) I opened dti_FA_alignstandard.nii.gz in fslview and using the JHU
> probabilistic atlas, I added the structure of interest then saved it
> to file.
>
> 5) I used the following command to make a binary mask (What is the
> rationale for using 25 as a threshold?)
> fslmaths tract_mask -thr 25 -bin binary_tract_mask
I assume 'tract_mask' is the atlas-derived mask that you saved out of
FSLView? Yes, thresholding at a probability of 25% is just an
arbitrary population-average level to threshold at, generally seems
reasonable. You can obviously go lower or higher if you want to be
more liberal or conservative, and that mostly depends on how confident
you are in the registration between your data and standard space.
> 6) Here's where I'm unsure of what to do. Before I had used the
> following commands to get the average FA of a mask:
> avwmaths_32R binary_tract_mask -mul dti_FA_alignstandard FA_tract
> avwstats FA_tract -M
> Now those commands don't exist, so I'm not sure how to use the
> fslmaths command to get the same result. Any thoughts?
Replace avwmaths_32R with fslmaths (and add '-odt float' to the end of
the line to make sure it saves out in floating-point format - that's
the equivalent of having the '_32R' in there before). Note that a
danger with this approach is that the average that you get out doesn't
just reflect the average FA within the tract, but also the quality of
the alignment and intersection between your standard-space-derived
mask and that tract in your subject - so unless you make the mask very
conservative, may include lots of non-tract FA values in the
averaging. For this you might consider either getting more
sophisticated with the way you pull out information from the
tractography, or use TBSS.
> Also, is there any way to pull out the individual labels of the JHU
> white matter label atlas? If I add one structure, it seems like it
> adds the whole atlas.
I'm confused - this works fine for individual structures for me using
both JHU atlases.
Cheers.
>
>
> Thanks so much for your help!
>
> Kristen
>
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library on behalf of Steve Smith
> Sent: Sat 2/23/2008 2:12 AM
> To: [log in to unmask]
> Subject: Re: [FSL] DTI-based White Matter Atlas
>
> Hi,
>
> Yes, you can do this; it's particularly easy if you have run the
> registration to highres and standard space in the FDT GUI.
>
> You can then go into FSLView, turn on atlas viewing, open
> 'structures', select the atlas and the structure of interest, press
> '+' to move that structure permanently into the image list, and then
> save that image to file. In general this will be a probabilistic mask
> so you may want to threshold it using fslmaths before using it as a
> mask in tracking, e.g.
>
> fslmaths thalamus_mask -thr 25 -bin binary_thalamus_mask
>
> Cheers.
>
>
> On 21 Feb 2008, at 12:34, Lindgren, Kristen, Ann wrote:
>
>> Thanks for the response, Steve. Really what I'd like to do is
>> directly identify a particular tract on the FA maps as opposed to
>> relying on tractography to identify it. Right now I'm getting
>> results by identifying two seeds in anatomical space and connecting
>> them, and I'm getting good results on some subjects but sometimes no
>> track on others. I thought the atlas was something like the
>> freesurfer atlas but for DTI data where I could create a label and
>> use that as a mask to get an average FA. I will look more carefully
>> at the atlas and see if it maybe would inform me of how to identify
>> my seed points in my FA maps for tractography. Thanks!
>>
>> Kristen
>>
>>
>> -----Original Message-----
>> From: FSL - FMRIB's Software Library on behalf of Steve Smith
>> Sent: Thu 2/21/2008 2:45 AM
>> To: [log in to unmask]
>> Subject: Re: [FSL] DTI-based White Matter Atlas
>>
>> Hi - the atlases are there as a guide for what structures are where -
>> but they may not exactly related to a given new individual's scan.
>> Using the masks directly in the context of TBSS results makes some
>> sense, but for tractography, if you have already got your tracts,
>> this
>> implied that you chose your seed/target voxels/masks carefully and so
>> presumably already have a strong idea of what tracts you are looking
>> for? So I'm not quite sure how you are wanting to then use the
>> atlases?
>>
>> Cheers.
>>
>>
>> On 19 Feb 2008, at 21:18, Lindgren, Kristen, Ann wrote:
>>
>>> Hello. I am interested in using the JHU DTI-based white-matter
>>> atlases
>>> to obtain the average FA values of various tracts. I've created
>>> masks
>>> and gotten average FAs from our DTI using tractography, but I'm
>>> wondering if there is documentation on how to use the available
>>> atlases
>>> to do this. Thanks and hopefully I explained that clearly!
>>>
>>> Kristen
>>>
>>
>>
>> ---------------------------------------------------------------------------
>> Stephen M. Smith, Professor of Biomedical Engineering
>> Associate Director, Oxford University FMRIB Centre
>>
>> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>> +44 (0) 1865 222726 (fax 222717)
>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>> ---------------------------------------------------------------------------
>>
>
>
> ---------------------------------------------------------------------------
> Stephen M. Smith, Professor of Biomedical Engineering
> Associate Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ---------------------------------------------------------------------------
>
---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------
