[Mristudio-users] fiber tracking capillary phantom

[susumu]

Hi Usha,
 
Because we have been using DtiStudio for many fixed mouse brains and fiber
tracking results can correctly follow the small convoluted paths of their
white matter tracts, I think the algorithm should be working without, at
least, a large amount of inaccuracy.
 
DtiStudio does not keep tracking results if they are only few pixel long.
That is why you sometimes don't see tracking results.
 
For your data, the first thing you want to check is the SNR. Please look at
the color map of your capilary phantom. Because the capilary is aligned
along Z, they should look blue. If they look noisy and not pure blue, then
SNR could be the culprit. Please also look at your vector image. You can
place your cursor in the capilary, right click, and see pixel values of
neiboring pixels. In this way, you can see FA values and vector values and
understand which threshold came in to prevent you from seeing long fibers.
With FA around 0.2, you need to have very high SNR to perform long-distance
tracking.
 
Regards,
 
Susumu

  _____  

From: mristudio-users-bounces at mristudio.org
[mailto:mristudio-users-bounces at mristudio.org] On Behalf Of Usha Sinha
Sent: Monday, February 09, 2009 2:38 PM
To: mristudio-users at mristudio.org
Subject: [Mristudio-users] fiber tracking capillary phantom


Hi 
I have been using DTIStudio for a while for tracking fibers in muscle.
Muscle fiber FA values are lower than white matter ( in the range 0.2-0.4).
I am currently evaluating with a capillary phantom with an FA of 0.20.  This
is a straight 5 cm capillary imaged with the capillary axis along the
z-axis.
I am trying to check the SNR conditions, as well geometry parameters that
provide an accurate estimate of the capillary length.  Slice by slice
evaluation of the capillary region clearly gives a value of +_2 degrees to
the z-axis for the leading eigenvector.  So our SNR is sufficient to give
the correct lead eigenvector orientation but the tracking results are
surprising:
A: Tracking condition:  FA threshold: 0.15, angle threshold: 10 degree
1) if a seed point of 1 voxel is used, the fibers are either short or there
are often no fibers tracked from the seed point (even though seed point
and neighbor voxels have values of FA above threshold) 
2) if the entire capillary array is selected,  then there are some fibers
yet average length of fibers is 2 cm (and not 5 cm as expected).  I guess
the difference between 1 and 2 is that in (2) several multiple connections
are found by the 'brute force' methods. 

B: Tracking condition:  FA threshold: 0.15, angle threshold: 70 degree
The number of fibers are many more for this condition still fiber length
does not reach 5 cm average (more like 3 cm)!
 
My questions::
1) Why are the fibers not tracked for 10 degree angle threshold?  These are
straight capillaries with angle differences only arising from noise (which
our measurements show are within 2-3 degrees)
2) Examining fibers from a single seed point, we see that fibers are not
entirely parallel to the capillary axis; rather short segments and curvature
of fiber is seen.
3) Is there a restriction on how anisotropic the slice voxel can be:  we
have in-plane of 0.3125x0.3125 and thickness of 5mm--- however capillary
axis is along the slice axis.  I wonder of this affects the interpolation
needed for FACT
 
This has  been of great concern-- please do send me some feedback
usha sinha
usinha at sciences.sdsu.edu
usinhaster at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20090216/a022202f/attachment.html