Hi,
On 4 Oct 2007, at 08:15, Agnieszka Burzynska wrote:
> Hi,
> Sorry for late response, but I was getting a visa in poland:)
>
> I traced back what happened and obviously 3 registartions did not
> work, but
> this I could see only after running tbss_3_postreg and slicesdir
> *.nii. I
> simply re-run the tbss_2_reg for these 3 subjects and was lucky
> they looked
> finally ok! Ufff...and the tbss_3 looked also reasonable. 3 questions
> remain:
> -how can I be sure that all other registrations are ok if running
> the same
> registration script twice gives different results?
In a sense, it doesn't (the program is deterministic) - if it
succeeds, then the results will always be the same - if it fails once
and succeeds a different time, then something must have gone wrong on
the computer before (e.g. running out of disk space, quota, RAM,
etc.). So if it all looks ok, then that's fine.
However you should always evaluate the all_FA image in FSLView, as
described in the TBSS manual; this is a very easy way of getting a
quick check that all the subjects were registered reasonably and in
line with the skeleton.
> -is it normal that the registered FA brain when overlaid with
> MNI152 does
> not overlap with it to the very edges of the MNI152 cortex?(white
> matter is
> well aligned, but the whole registered brain looks a bit smaller)
Yes, this is normal as the FA values fall off so much once you reach
grey matter. To get a feeling for what's "normal" you could overlay
the standard space FMRIB58_FA image on top of the MNI152 in FSLView.
> -how to adjust cluster-forming t threshold: is 1.5 also ok for my 35
> subjects?
See previous emails on this - I'm afraid there's no good answer to
this; that's one of the weaknesses of cluster-based thresholding. 1.5
is valid (in the sense that randomise will end up giving you valid
corrected p-values in the maxc image, regardless of the cluster-
forming threshold), but may not be "optimal" for your data.
Cheers, Steve.
> Thank you!
> Best,
> Aga
>
>
>
>
> On 03.10.2007 10:15 Uhr, "Steve Smith" <[log in to unmask]> wrote:
>
>> Hi - if the initial image orientations are not axial or the images
>> are a cut-down FOV then sometimes the IRTK registration can fail.
>> Might this explain it?
>>
>> The tbss_2_reg.e12345.6 etc files are just log files so you don't
>> need to copy them.
>>
>> What happens if you run
>> fslinfo all_FA
>> fslstats all_FA -r -R
>> in the stats directory? Does it have the right number of timepoints?
>>
>> Cheers.
>>
>>
>>
>> On 1 Oct 2007, at 10:20, Agnieszka Burzynska wrote:
>>
>>>
>>> Hi,
>>> A little follow-up:
>>> I checked all my data from the very beginning: the FA maps looks ok
>>> as well as scaled ANALYZE files after tbss_1_preproc. 2
>>> registrations look bad after tbss_2_reg: on one there is nothing
>>> and on the second looks as if half of brain is cut off. I am
>>> trying to repeat the registration for these 2 now, hoping this will
>>> solve the problem. Do you know why so many registrations work well
>>> but for some it doesnt work? Is there anything else that could have
>>> gone wrong?
>>>
>>> Thank you!
>>> Aga
>>
>>
>> ---------------------------------------------------------------------
>> ---
>> ---
>> 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
------------------------------------------------------------------------
---
|