Hi Matt,
The primary concern was that there's minor techinical issues in making
the symmetrised skeleton in the first place, and then possible subtle
gotchas wrt projecting the data onto this. My script will (if _I've_
got it right....) make sure that these are not problems. However,
I would not think that these relatively minor issues would likely
cause (e.g.) large areas of spurious (false positive) results. It's
possible that the final details of how one then flips the data and
runs randomise might cause problems - but at that stage hopefully the
possible dangers should be pretty obvious....for example we will
probably recommend the approach of flipping the data, subtracting, and
then masking out one half of the brain, and running that inside
randomise - I don't know whether keeping both halves could induce
false positives......?
Cheers
On 19 May 2009, at 04:20, Matt Glasser wrote:
> Hi Steve,
>
> I read this note with some concern, given that a couple of my
> colleagues are
> using TBSS to evaluate hemispheric FA asymmetries. In looking over
> their
> results, I think they have found too many significant voxels
> (including a
> large number of the highest possible significance). What biases one
> might
> expect if they used the publically released TBSS scripts to perform a
> hemispheric FA asymmetry study?
>
> Thanks,
>
> Matt.
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf
> Of Steve Smith
> Sent: Thursday, May 14, 2009 11:23 AM
> To: [log in to unmask]
> Subject: Re: [FSL] Symmetrical TBSS skeleton
>
> Hi,
>
> Yes - when you want "tbss_skeleton" to use a specific skeleton rather
> than one derived from the 4D data you feed in, you need to use the -s
> flag (see the usage).
>
> Also, you need to explicitly force the skeleton to be symmetric and
> not just derive it from a symmetric mean FA image (the skeletonisation
> will not necessarily give an _exactly_ symmetric skeleton when fed a
> symmetric image).
>
> We have a script for doing asymmetry analyses which we will clean up
> and make available soon - hopefully next week.
>
> Cheers.
>
>
>
>
>> Hi,
>>
>> Just to correct my last post - what I get at the end of this
>> processing
>> pipeline is an 4d all_FA_skeleton file that IS symmetrical but does
>> not
>> quite fit my original "symmetrised" mean_FA_skeleton in that there
>> are
>> additional/missing voxels in places.
>>
>> Is this a reasonable method to achieve what I am after, or is there
>> an
>> alternative method.
>>
>> Thanks.
>>
>> Pete
>>
>
> On 11 May 2009, at 12:44, Pete Goodman wrote:
>
>> Dear All,
>>
>> I am trying to use TBSS to derive asymmetry indices. To do this I
>> have,
>>
>> 1) Made the FA template symmetrical by creating a flipped version
>> (fslswapdim applied) and averaging the flipped and unflipped versions
>> 2) Registered all data to new template using tbss scripts 1, 2 and 3
>> 3) Taken the mean FA skeleton and made this symmetrical using the
>> same
>> method as above
>> 4) Fed my all_FA data into tbss 4 together with the new symmetrical
>> skeleton
>> - here I was expecting voxels from each subject to be projected onto
>> the
>> symmetrical skeleton which I could then flip again and use
>> (unflipped minus
>> flipped) to derive AI.
>>
>> However, the all_FA_skeletonised output I get is not quite
>> symmetrical. The
>> deprojection step seems to have ignored the use of the new
>> symmetrical
>> skeleton, and instead used the old, unsymmetrical one.
>>
>> Any help, tips as to a better method would be welcome.
>>
>> Thanks.
>>
>> Pete
>>
>
>
> ---------------------------------------------------------------------------
> 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
---------------------------------------------------------------------------
|