Optimizing surface renderings

Questions about BrainVisa usage and installation

Moderators: riviere, denghien

gzr
Posts: 14
Joined: Sun Nov 02, 2008 3:18 pm

Optimizing surface renderings

Postby gzr » Wed May 12, 2010 7:36 pm

I'm processing a new set of brains and the surface renderings are not great (e.g., lots of indentations/dents, etc). I suspect it's due to the fact that the images are relatively old and not of the best quality. However, if I do want to try to tweak the processing settings to improve the results, is there any parameter/setting that you would recommend trying to change first?

(I realize the answer is probably, "No - you just have to play around with all the different options to see if anything helps"...but figured it was worth asking in case someone has recommendations since there are so many parameters to choose from :-) )

Thanks!

User avatar
riviere
Site Admin
Posts: 1358
Joined: Tue Jan 06, 2004 12:21 pm
Location: CEA NeuroSpin, Saint Aubin, France
Contact:

Re: Optimizing surface renderings

Postby riviere » Wed May 12, 2010 10:27 pm

Hi,

To make it short, the critical step is almost always the bias correction. I usually play with only 3 parameters:
- the correction algorithm variant, by switching between the two sub-processes of the bias correction step
- the field rigidity
- the sampling parameter
If it still doesn't work perfectly, it is still possible to "fix" the histogram analysis by hand, by editing the histo_analysis (it's a text file) and overriding values for grey and white matter grey level peaks and std deviations.

Denis

gzr
Posts: 14
Joined: Sun Nov 02, 2008 3:18 pm

Re: Optimizing surface renderings

Postby gzr » Fri May 14, 2010 4:17 am

Great - I'll give it a try. Thanks!!

gzr
Posts: 14
Joined: Sun Nov 02, 2008 3:18 pm

Re: Optimizing surface renderings

Postby gzr » Sat May 22, 2010 3:26 pm

I've been trying different things based on this thread and the "Error in Pipeline" thread. So far the alternate "Bias Correction" option looks the best (the alternate Histogram option yielded fatal errors on 2/5 brains, and the alternate Grey-white Interface option did not result in any changes).

The next thing I was going to try is adjusting the field rigidity and/or sampling parameter, but I was wondering if you could help me identify the amount by which I should change these. Since they are continuous variables, I feel like I would be choosing alternative values relatively arbitrarily (though, based on the documentation I think I should start by _increasing_ the field rigidity).

I'll attach a couple of sample images here, in case that is helpful. (I'm only concerned with the ventral surface so have attached relevant screenshots) Due to size limitations, I'll attach one to this message and a second to a new reply. For both images, the version produced using the default pipeline configuration is on the left, and the improved version from using the "Vip Bias Correction" option is on the right.

Thanks!!!
Attachments
Screen shot 2010-05-22 at 10.19.21 AM.png
Left hemisphere processed using (1) Default Pipeline configuration (left image), and (2) Vip Bias Correction
Screen shot 2010-05-22 at 10.19.21 AM.png (157.75 KiB) Viewed 3770 times

gzr
Posts: 14
Joined: Sun Nov 02, 2008 3:18 pm

Re: Optimizing surface renderings

Postby gzr » Sat May 22, 2010 3:32 pm

Here is the second attachment. For this one, the alternative Bias Correction option made a dramatic difference and seems to have resolved the initial processing errors
Attachments
Screen shot 2010-05-22 at 10.18.16 AM.png
Left hemisphere processed using (1) Default Pipeline configuration (left image), and (2) Vip Bias Correction
Screen shot 2010-05-22 at 10.18.16 AM.png (153.42 KiB) Viewed 3758 times

User avatar
Jean-Francois Mangin
Posts: 337
Joined: Mon Mar 01, 2004 10:24 am
Location: Neurospin, CEA, France
Contact:

Re: Optimizing surface renderings

Postby Jean-Francois Mangin » Wed Jun 02, 2010 2:20 pm

Sometime, the topological trick to get nice rendering with open folds fails.
You can force it removing the folds from the data directly using this set of commands:
I will try to put the option somewhere in the next releases...

VipSingleThreshold -i Lskeleton_subject.nii -t 19 -c b -o Lskel.nii
VipSingleThreshold -i Rskeleton_subject.nii -t 19 -c b -o Rskel.nii
VipExtEdge -i Lskel.nii -m 3Da -o Lskele.nii
VipExtEdge -i Rskel.nii -m 3Da -o Rskele.nii
VipMerge -i Lskel.nii -o Lskel.nii -m Lskele.nii -v 0
VipMerge -i Rskel.nii -o Rskel.nii -m Rskele.nii -v 0
VipMerge -i voronoi_subject.nii -m Lskel.nii -v 0 -o voronoi_subject.nii
VipMerge -i voronoi_subject.nii -m Rskel.nii -v 0 -o voronoi_subject.nii
rm Lskel.*
rm Rskel.*
rm Lskele.*
rm Rskele.*

gzr
Posts: 14
Joined: Sun Nov 02, 2008 3:18 pm

Re: Optimizing surface renderings

Postby gzr » Thu Jun 03, 2010 4:57 pm

I apologize for being naive, but I have not yet tried manually inputing any commands and am not really sure how to do that (I think that is what you are suggesting I do, correct?). When I try to open a "shell" the program crashes, so I'm hoping that's not how to do it. Do I need an additional program, like Matlab, to be able to do this?

User avatar
riviere
Site Admin
Posts: 1358
Joined: Tue Jan 06, 2004 12:21 pm
Location: CEA NeuroSpin, Saint Aubin, France
Contact:

Re: Optimizing surface renderings

Postby riviere » Fri Jun 04, 2010 1:08 am

Hi,
Commands are not based on Matlab, you don't need additional software.
You just need to open a terminal (or a DOS terminal, on Windows), and the programs are in the "bin" directory of BrainVisa package. There is a script to setup paths, also in the "bin" directory, called "cartopack.sh" or "cartopack.csh", or "cartopack.bat" depending on your shell type, which you may 'source' to make programs directrly available in the paths. This is explained in the README file in the root directory of the package.
When I try to open a "shell" the program crashes

You mean the "open shell" menu in BrainVisa ? Which operating system are you using ? Did you run brainvisa from a terminal or by clicking on it in a graphical environment ? The shell option only works when brainvisa is run from a terminal.
Denis

gzr
Posts: 14
Joined: Sun Nov 02, 2008 3:18 pm

Re: Optimizing surface renderings

Postby gzr » Sat Jun 05, 2010 3:36 pm

Ah, that makes sense. Thanks! I'll give it a try...


Return to “BrainVISA”

Who is online

Users browsing this forum: No registered users and 5 guests