Hi Nil,
I'm not sure if I completely understand want you want to visualize. If
you want the axes to be aligned with the image layers a simple reinit on
the dataset should do the trick, no? If you want to visualize the
principal directions of a plane rotated relative to the image layers you
indeed end up with many voxels that are only partially inside of the
image plane. Since the principal directions are at the moment simply
realized as a fiber bundle, this results in the problem you described
with the partially visualized directions. I guess to solve this, MITK
would need a dedicated data structure for visualizing vector fields that
can handle the reslicing necessary to obtain a "virtual" plane that is
not parallel to the actual image layers. I hope I expressed that
correctly :) Currently we do not implement this if I'm not mistaken, but
in general it should be possible.
Cheers
Peter
On 16.12.2014 14:46, Nil Goyette wrote:
Hi all, Hi Peter Neher,
We are trying to see the tensor direction, like you do in MITK
Diffusion in the "Tensors" plugin in "Start Tensor Principal Direction
Extraction". Thanks to you, the work is all done :) This being said,
we have one problem. It works perfectly on an image without rotation
(translation and scaling is ok), but we have trouble with rotated
images. http://imgur.com/UR9JnKO
- We can see the lower and/or upper slice
- The tensors are not "in" the voxel, but floating in its
neighborhood. It can be impossible to say which voxel a tensor belongs to.
- The crosshair is not in the center of a voxel. (Not a bug I guess
because the image is rotated)
- The image navigator (not shown in screenshot) and step size are
wrong. (Idem???)
I'm pretty sure I understand what's going on. On
http://imgur.com/gaJhm7U, I printed many tensors per voxel but only on
one slice. The image is rotated but the axes aren't, so we end up
crossing a lot of slices. Since we are not exactly in the center of a
layer, we can see the end of a tensor on slice i and the start of a
tensor on slice i + 1.
I tried fixing the problem by rotating the axes as best as I can so
they "follow" the image. http://imgur.com/yVLKj5c If you place the
crosshair on the right voxel, it centers some tensors. It's hard to
find the "right" pixel and we still have the layer problem. I'm not
sure how MITK works on the inside, but it seems that we are stuck with
interpolation and bad step size if the image is rotated, even if you
try to rotate the axes so it fits.
What are your thoughts on this? Is tensor visualization on rotated
images an impossible idea in MITK? Is the only way to succeed is to
remove the rotation in the transformation matrix? Thanks for your time.
--
Logo Imeka <http://imeka.ca/> Nil Goyette, M.Sc.
www.imeka.ca <http://imeka.ca/>
--
Dr. sc. hum. Peter F. Neher
Division Medical and Biological Informatics
Junior Group Medical Image Computing (MIC)
German Cancer Research Center (DKFZ)
Im Neuenheimer Feld 280
69120 Heidelberg
Germany
phone: +49 6221 42-3552
fax: +49 6221 42-2345
[email protected]
http://www.dkfz.de/de/mbi/people/Peter_Neher.html
Confidentiality Note:
This message is intended only for the use of the named recipient(s) and may
obtain confidential and/or privileged information. If you are not the intended
recipient, please contact the sender and delete the message. Any unauthorized
use of the information contained in this message is prohibited.
------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk
_______________________________________________
mitk-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mitk-users
