[Mristudio-users] Landmarker - Reorientation questions

[susumu]

Hi Paul,

 

The landmark-based registration simply follows your landmarks. So, if you
want to modify the outcome, you need to modify (or add more) landmarks.

 

You can also test automated volume registration by AIR.

 

Either way, you can save the transformation matrix and apply it to fiber
tracking results.

 

There is one important limitation in our current software when you normalize
the tracking results.

For the normalization, you have to convert the streamline information
(DtiStudio-specific fiber data format) into pixel-by-pixel image format.

 

For example, suppose your tracking generated 10 fibers with average 50-pixel
long. Then your native streamline information contains [x,y,z] information
for the 10 streamlines saved sequentially.

 

This streamline information is not easy to transform. I believe Luke in my
lab has an IDL code to do this, but our MriStudio family can not do it.
Please contact khua at jhu.edu if you want the IDL code.

 

You can convert this streamline information to 1/0 pixel-by-pixel image
format. For example, if your data is 256x256x50, the pixels that contain the
tracking result are 1 and do not contain are 0. You can save and load this
info as an image file. Then you can transform (normalize) this file as if
one of your MRI contrasts. The problem of this approach is, you can not
retrieve streamline info once it was converted to the 1/0 binary image.

 

You should think why you need to retain the streamline information after
brain registration (normalization). For example, you can create all
necessary information in a native space such as connectivity maps and
transform the resultant map.

 

Susumu

 

  _____  

From: mristudio-users-bounces at mristudio.org
[mailto:mristudio-users-bounces at mristudio.org] On Behalf Of Paul Beach
Sent: Thursday, March 05, 2009 10:52 AM
To: mristudio-users at mristudio.org
Subject: [Mristudio-users] Landmarker - Reorientation questions

 

Greetings everyone,

I have been attempting to use Landmarker to re-orient subject brains to a
template provided by the developers (we successfully changed the template to
our dimensions), and I have several questions regarding rigid
transformations.

I have successfully transformed subjects using several anatomical landmarks
(AC/PC, 1st and 2nd mid sagital, most anterior/posterior/superior/

inferior/left/right).  However, subject brains are still usually a bit
slanted in certain views, post-transformation.  Is there any advice as to
how to correct for this occuring?  Is it likely a landmark placement issue?

Also, once the transformations are made, I have been attempting to apply the
saved landmarks for the subjects to the fiber tracking data (say for a
bilateral cingulum) that we have already produced for that subject using MRI
studio.  However, we are a bit stumped in attempting to find a way to take
that reoriented fiber tract data and view it 3D on MRI studio without having
to redo all the ROI placement, etc.  Any advice for this?

I have also been experimenting with using the rotation/shift image buttons
to change subject images (I actually used this to fix a problem with the
template, as when we re-sliced it to our dimensions it was shifted
off-center axially/coronally to the user-right).  Would using these buttons
affect the results of a rigid transformation?

Also, I was curious regarding the general algorithm for the rigid
transformation.  Is it a similar one to AFNI's manual Talairaching(?) or
does Landmarker simply re-orient the brain in a way such as to minimize the
variance in distance between all the placed landmarks?

Many thanks for your assistance,

Paul




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20090305/eb3bf535/attachment.html