Hi,
On 23 Mar 2007, at 19:02, Antonios - Constantine wrote:
> Dear fsl users,
> I used siena and sienax to process longitudinal and cross sectional
> 25 data
> sets. Each data set consists of 2 T1 scans at 2 different
> intervals. Both
> scans are of different FOV and quality. The first time interval's
> scan is of
> 256x256x124voxels and 1.093x10.93x1.5mm voxel dimensions while the
> other is
> of 256x256x50 and 1.093x1.093x 2.5mm.
> I had to process 10 data sets of Controls (with scans at 2different
> intervals ) and 15data sets of Parkinson Disease patients (with
> scans at 2
> different intervals also) and, correct me if i'm wrong, i used the
> same
> parameters settings for all the data sets in siena and sienax
> respectively,
> in order to have an objective comparison of the results.
>
> In almost all of the data i had very good results except from 3
> data sets.
> In these data sets the Bet tool failed totally, hence both siena
> and sienax
> gave bad results and siena printed in 2 out of 3 datasets a warning:
> “Probably failed consistency, check for standard space registrations”
>
> I have the following questions:
>
> 1)what's the meaning of the siena's warning output? Cause i didn't
> find any
> info about that in fsl's web page /lectures.
This means that the various registrations were not consistent between
the two images, which means that at least one of them failed. Your
two input images are quite different from each other so the value of
running SIENA between them is rather questionable anyway I'm afraid.
However....
In the example you sent the two images registered to each other ok,
but the second image didn't register well to standard space - this
may be because BET isn't working on this image with these options. I
suggest that you use bet_robust (script attached) instead of bet in
the siena script (take a local copy of $FSLDIR/bin/siena and change
the bet calls near the start) and remove the BET options that you
added before. You could probably turn off the -m option then as well.
This then runs through SIENA ok, but the images are also somewhat
distorted relative to each other - I really don't think it's a great
idea to use such different pairings in SIENA!
> 2)why i didn't have these problems in siena with the other data
> sets , since
> all of them are of different quality and FOV ?
Presumably because FLIRT succeeded in those cases but not the more
problematic ones.
> 3)If i change sienax parameters then i'll probably find a set that
> will help
> Bet to succeed but if i'll do that then how can i compare the
> results of all
> my data sets since the comparison won't be objective??
I'm afraid you can't really use SIENAX to compare between very
different acquisitions, such as you have here. The segmentation
output volumes almost certainly won't be comparable between
timepoints, though you could just use the first OR second timepoints
to compare volumes across subjects. In this case, if you hand-tune
BET to work well across subjects such that it looks good in each, you
are probably ok - though again you may find that bet_robust works
better anyway without needing hand tuning.
> 4)Is it wrong that i used siena to make longitudinal analysis in
> data that
> have different FOV, without reducing the FOV of the “bigger” image
> till it
> reaches the “smaller” one?
Each pairwise analysis is ok - but you need to be careful that the
results are comparable across subjects.
> 5)I used siena in all my data with the following set up: siena A B
> “parameters settings” where A is the scan with the smaller FOV and
> less
> resolution (2.5mm ) and B the scan with the bigger FOV and higher
> resolution
> (1.5mm). If i change the sequence of the scans in siena :siena B A
> “parameters settings” there is a difference in the PBVC. why is that?
In general this doesn't happen - in this case it's probably related
to the standard space registration problems.
> 6)In some results i noticed that after segmentations the eyes were not
> excluded. does this induce a serious error in siena/sienax results?
> and if
> yes how can i avoid that without changing the siena/sienax
> parameters which
> will make the final results not objective for comparison with the
> other
> results of my data??
Hopefully these issues will be helped by the above.
Cheers, Steve.
>
> Thank you once again for your valuable support
> Antonios-Constantine Thanellas
>
>
> P.S. I've uploaded one of the data sets where i had these problems
> with the
> output of the siena script on these data. They are control data
> from the
> same subject in different intervals. The ref ID is :377949
------------------------------------------------------------------------
---
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
------------------------------------------------------------------------
---
|