Rigid registration with MI dialog box

Questions about BrainVisa usage and installation

Moderators: denghien, riviere

Post Reply
User avatar
rvboutin
Posts: 36
Joined: Tue Feb 09, 2010 12:53 pm

Rigid registration with MI dialog box

Post by rvboutin »

Hi,
Few naive questions about the Rigid registration with MI dialog box, what are the reference_threshold, source_threshold, reference_reduction_factor and error_Epsilon parameters referring to? I manage to co-register preclinical images with this tool but it takes ages to compute... I was wondering if altering the default values for these parameters might help? but to do so, I'd like to understand what they are for...
Cheers,
Rv
__________________________________
Hervé BOUTIN, PhD
Senior Research Scientist
University of Manchester
Faculty of Biology, Medicine and Health
email: herve.boutin@manchester.ac.uk
__________________________________
denghien
Posts: 209
Joined: Mon Apr 05, 2004 1:35 pm
Location: IFR49 / SHFJ Orsay
Contact:

Re: Rigid registration with MI dialog box

Post by denghien »

Hi,

You could find some information on this part of documentation (Aims command line):
http://brainvisa.info/doc/aimsdata-4.3/ ... 08s02.html

In fact, Brainvisa interfaces the AimsMIRregister command line, which is usable in a terminal too.
If you launch : AimsMIRegister -h, you will get some information about options.
When a command is interfaced by BrainVISA, some options can be hidden to make easier to use.

Here the link between brainvisa and command line names :
"reference_threshold" = [ --threshref | --seuilref <FLOAT> ] : relative thresh applied prior to grav cent estimation
"source_threshold" = [ --threshtest | --seuiltest <FLOAT> ] : relative thresh applied prior to grav cent estimation
"reference_reduction_factor" = [ --refstartpyr <S32> ] level of the multiresolution optimization: start resolution level for ref
"error_Epsilon" = [ --error <FLOAT> ] : tolerance on results [default=0.01]

I tried with 2 T1MRI. It tooks 20 minutes. What kind of images do you use ?
Have you reached a result ?
If you want I could try with your images.

Isa
User avatar
rvboutin
Posts: 36
Joined: Tue Feb 09, 2010 12:53 pm

Re: Rigid registration with MI dialog box

Post by rvboutin »

Hi,
OK thanks.
Could you explain more about what "level of the multiresolution optimization: start resolution level for ref" and "error_Epsilon = [ --error <FLOAT> ] : tolerance on results [default=0.01]" means...
I did try to co-register a T2 MRI or T2 MRI template with a CT image, but that did not work very well, so I am using mask instead, and that works fairly well. I am just trying to get it faster. So I am wondering if altering these parameters might make the process faster.
Cheers,
Rv
__________________________________
Hervé BOUTIN, PhD
Senior Research Scientist
University of Manchester
Faculty of Biology, Medicine and Health
email: herve.boutin@manchester.ac.uk
__________________________________
denghien
Posts: 209
Joined: Mon Apr 05, 2004 1:35 pm
Location: IFR49 / SHFJ Orsay
Contact:

Re: Rigid registration with MI dialog box

Post by denghien »

Hi,

If you use the "reduction_reference_factor" (it works on reference image), you will damage the image.
For example, if you choose the value "1", then 1 voxel out of 2 in the 3 directions will be damaged.
It is interesting if images doesn't have the same modality in particular with intensity differences.
It seems the case, you could try with a value "2" or "3" too.
(you could do the same thing on source image, if you work with command lines)

Another possibility: you could try by changing the initial mode "ini_with_gravity_center", by default, the registration is initialized by using the gravity center.
In this case, you could initialize the registration with a translation by using "initial_translation_x' and so on.
It is interesting when images are far from each other.

Isa
Post Reply