Dear Donna
thanks for your kindly help again.
please see messages within the letter you replied.
Po-he
Message: 3
Date: Thu, 18 Nov 2004 12:37:36 -0600
From: Donna Hanlon <[EMAIL PROTECTED]>
Subject: Re: [caret-users] Re: cc bridge, peaks, etc.
To: "Caret, SureFit, and SuMS software users"
<caret-users@brainvis.wustl.edu>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset=UTF-8; format=flowed
On 11/17/2004 11:52 PM, pohetsn wrote:
Dear Donna
Thanx for your detail explanation.
No flip is need to the volume.
I'm very interested in learning more about the CC bridge problem.
But in your latest uploaded volume, you've cropped well to the left of
the midline, so there is no contralateral hemisphere left. Since the
details of how to remove the bridge are specific to your volume, I can't
address this issue with your latest dataset.
The volumes I uploaded are only for you to see which surface lies better
on layer4.
Is it realted to the ability that surefit can only do half-hemisphere
segmentation?
What does half-hemisphere segmentation mean? I thought you meant that
SureFit can only segment one hemisphere at a time, but then your
uploaded volume shows an unusual cropping -- well lateral of the
midline. Is this really what you want, or do you want the full
hemisphere?
The CC bridge problem occurs when the volume is rotated significantly
off midline. SureFit doesn't have any built-in spatial normalization
routine or control points for specifying where the midline is located;
it uses landmarks relative the AC for separating the hemispheres.
You are right. I mean one hemisphere at a time.
And the lateral side is not midline.
The following is my workflow for future segmentation.
would you please give me some advice to it?
1. file format transformation from .img to .mnc
2. resample to 1mm cubic
3. my goal is to do half-hemisphere segmentation.
While initial testing, I first crop a volume that I interested most
to see which GM and WM peak value lead to best surface boundary.
(I uploaded chengcc2.tar.gz to your web.
There are 3 segmentation with different WM peak value.
Would you please tell me which segmentation has the best boundary
fitted to layer 4?
Because I read some anatomy textbook. Layer 4 sometimes is very
close to white matter, but sometimes not.)
In the case you uploaded, I would have chosen GMpeak=55, WMpeak=74.
Your 1st_test_segment.jpg looked closest to these settings. The others
looked too biased toward gray matter.
Thanks a lot.
By the way, I'd like to confirm that whether SureFit aims for a
segmentation
whose boundary runs approximately "midway of inner bourndary and outher
boundary"
or "layer 4"?
4. apply this GM,WM peak values to the whole hemisphere and run SureFit.
5. correct all handles on desired hemisphere, neglecting those on the
other hemisphere or others.
6. Run Surefit to segment the fixed-handles segment again to see
whether only little handles are left.
7. if yes, fix the rest, and remove any part that is not on the
desired hemisphere.
8. use "Generate surface" to construct ideal surface. (Although this
surface won't have corresponding handle-painted surface.)
5: Before I start handle-fixing, I usually get rid of big "appendages"
like your CC bridge (which I'll email detailed steps for removing
later). This is because the appendages contain many handles and getting
rid of them reduces the handle count significantly. Sometimes, the
appendage isn't a CC bridge, but rather residual eye fat or optic nerve
that failed to disconnect during segmentation. After removing big
appendages, I then fix handles I can see on the inflated surface.
6, 7, and 8: You can Run SureFit a second time with these settings:
* use segmentation in vol2
* fill ventricles no
* correct errors yes
* identify sulci no
But I usually don't bother to do this, because it takes a long time and
usually isn't worth it (i.e., you often end up having to fix the
residual handles yourself anyway). Usually, I fix whatever handles I
can find, and then run SureFit like this:
* use segmentation in vol2
* fill ventricles no
* correct errors no
* Prepare for Flattening
* identify sulci yes
Then, I proceed to flattening and registration in Caret.
I see, so I must be 100% sure about all handles are fixed at once.
by the way,
it is fine to have all the ventricles filled. just curious why some
are filled but other are not.
If you are doing a full hemisphere, I recommend selecting Fill
Ventricles yes the first time you Run SureFit (i.e., after you select
your peaks). Then, it will try to fill the whole ventricle, but
sometimes it fails -- usually when the hemisphere is rotated
significantly off AC-PC line.
and how do i see smooth image again in the SureFit main window after
enlarging it.
After installing the patch, the enlarged and unsmooth images are quite
hard to read.
Pressing any of the view buttons (i.e., M,L,A,P,D,V and R) should
"unzoom" the surface. The R button stands for Reset. Hitting 'r' in
the surface window will reset the scale, but not the rotation. I hope
I'm understanding your question correctly.
Sorry for unclear expression.
The "SureFit main window" i meant is not the surface displying window.
I meant the window displays sMRI slices which segmentation can overlay on
it.
thank you very much for your help again
Sincerely,
Po-he
------------------------------
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
End of caret-users Digest, Vol 16, Issue 8
******************************************
