Print

Print
Dear Mark and Andreas, 

I really appreciate you taking a look at my images.  I did check the images in FSLView and they look fine on my end.  I have now run sienax using the default settings on a number of subjects.  This fixes the orientation problem, which is great.  For the brain-clipping problem at the FLIRT stage, it looks better for some subjects, but not completely.  It is still clipping some of the brain, as the standard brain mask is still smaller than the subject's brain.  Any thoughts on how to ensure that the standard brain mask encompasses the total individual brain?  

Here is the report: https://www.dropbox.com/s/mgxs5rwmyigjvnz/subject3report.pdf
You can see that it is still clipping a little of the anterior portion of the brain in the axial view, and a little dorsal part in the sagittal and coronal views. 
Here is the image for this subject:  https://www.dropbox.com/s/flzghqd729zfqel/vc30451_mpr_n1_111_t88.nii
hopefully the file isn't corrupted!  

Thanks so much again, 
-Deanna


On May 25, 2012, at 8:30 AM, Mark Jenkinson wrote:

Dear Deanna,

Andreas Bartsch said that he checked your images from the
download and they were OK.  So it is possible that my download
was corrupt (as I am not in the lab) so ignore my previous
message - although it is always good to check your images - as
it seems that what I saw is unlikely to be a problem with your data.

All the best,
Mark


On 25 May 2012, at 18:26, Mark Jenkinson wrote:

Dear Deanna,

I have just had a look at these images.
Both are completely corrupted in the conversion to NIFTI.
What I see is a collage of pieces of the brain in the wrong place
(some wrap-around but some much worse and arbitrary).
This needs to be fixed in the conversion to NIFTI.  Please have
a look at the data before doing any analysis.  If you are using a
viewer other than FSLView and it looks OK then that isn't enough
and you'll need to also check in FSLView (where I needed to reset
the max display range to 2000).  Only if it looks OK in FSLView
should you proceed with any other analysis. The problem with these
images is very obvious and should be immediately apparent when
you look at them in FSLView.

All the best,
Mark



On 17 May 2012, at 05:21, Deanna Greene wrote:

Thanks, yes something strange is going on.  If anyone else is able to take a look at my original structural images, I would really appreciate it.

Here is the structural image from the subject who is being oriented incorrectly at the FLIRT stage:   https://www.dropbox.com/s/pcv6mcqmo46aaub/sub1_mpr_n1_111_t88.nii
and just for reference, here is my sienax html output from this subject:  https://www.dropbox.com/s/fhp65rbcn60nf0w/sub1.pdf

Here is the structural image from the subject whose brain is getting clipped at the FLIRT stage:  https://www.dropbox.com/s/5qf0cbil1b4dbq2/sub2_mpr_n1_111_t88.nii
and again for reference, here is my sienax html output from this subject:  https://www.dropbox.com/s/ca0n3ygprrek8ce/sub2.pdf

Thank you to anyone who could take a look at this while Steve is out of town!
-Deanna


On May 16, 2012, at 2:02 PM, Stephen Smith wrote:

Hi - in both cases it looks like there's a problem with FLIRT, which I don't understand because if BET has worked well (looks like it has) and the image is in the right orientation to start with it is VERY rare for FLIRT not to work well and robustly.  Is it possible that maybe your voxel sizes are not set correctly in the header?

I'm afraid I'm now going to be away from email for the next 3 weeks travelling around China - so can't follow up on this - but maybe you should put your original structural images onto DropBox and someone might be able to look into the FLIRT question.

Cheers, Steve.


On 17 May 2012, at 03:43, Deanna Greene wrote:

Thank you for the quick reply.  

I tried to reorient the image, but it seems that it was already in the same orientation as the MNI152 template.  So when I ran sienax again (just to try), the standard space registration results are still oriented incorrectly.  Here is an image of the html output:  https://www.dropbox.com/s/fhp65rbcn60nf0w/sub1.pdf

For my second issue, here is an image of the html output:  https://www.dropbox.com/s/ca0n3ygprrek8ce/sub2.pdf
I am particularly worried about the anterior part of the brain that is being excluded from the whole brain segmentation, as can be seen in the sagittal and axial views.  How can I make sure the standard brain mask encompasses the entire brain?  

Thanks again!
-Deanna

On May 15, 2012, at 9:27 PM, Stephen Smith wrote:

Hi

On 16 May 2012, at 10:20, Deanna Greene wrote:

Dear FSL experts,

I am trying to normalize volumes by ICV in order to make group comparisons.  After looking through many threads on calculating ICV, it seems that using the VSCALING value from sienax as a normalization factor is the best approach.  So, I have run sienax on my subjects, but I am noticing problems with the output from some subjects.  

1.  For a few subjects, it appears that the individual structural scans are going into standard space in the wrong orientation.  The BET results look good (I think), but the FLIRT standard space registration results are oriented wrong.

The simplest thing to do is to reorient them by hand to have the right general orientation as the MNI152 template - you probably can run the same reorient command on all images.

If your header has enough orientation image, you can use fslreorient2std

2.  The other issue is that for about half of my subjects, the standard brain mask is smaller than the BET-derived brain mask (and smaller than the subject's brain).  Therefore, when I look at the "Field of view and standard space masking" in the html output, there is a ring of blue (from the BET-derived brain mask) showing the part of the BET mask that does not overlap with the standard space mask.  The result of this seems to be that the whole-brain segmentation cuts off part of the brain, so I imagine that the size of the brain is being underestimated, which worries me when using the VSCALING value.  

This sounds like the brain extraction of registration is not working well - without knowing more details, i couldn't say which.

You can put images on your website, dropbox, etc for people to see.

Cheers.




Does anyone know how I can fix these problems?  Also, if someone could tell me how to attach the images from the html output without exceeding the size limits for posts, that would be very helpful, as I think seeing the output would help convey the issues.

Thank you very much!
-Deanna



---------------------------------------------------------------------------
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
---------------------------------------------------------------------------