[Mrtrix-discussion] fiber statistics-MD, AD, RD

[J-Donald Tournier]

Hi Matt,

You can actually get all the values you need within MRtrix - no need for
Excel:

$ tensor_metric dt.mif -adc MD.mif -fa FA.mif -num 1,2,3 -value D[1:3].mif
$ mradd D2.mif D3.mif - | mrconvert - -scale 0.5 RD.mif

At this point, you'd have the MD & FA maps, and each eigenvalue as a
separate map (D1.mif -> D3.mif). D1.mif is the AD, and the second command
gets you RD explicitly as (D2+D3)/2.

After that, it looks like you're using the first eigenvalue to determine
the mask for the response function estimation. I'm not sure whether you
have a particular reason to do that, but the general recommendation (at
least in this version of MRtrix) is to use the FA map, thresholded at
around 0.7 (ideally you want to end up with ~200-300 voxels in the mask,
placed in regions you'd expect to contain predominantly single fibre
populations). I notice you say you followed the instructions in the
documentation to get the FA, but nowhere is the FA computed in your
commands...

Most other commands look fine until your threshold command - really you
want to set an explicit threshold value here using the -abs option,
otherwise you'll get a variable threshold depending on exactly how the
automatic threshold determination decided to work on that particular image,
which wouldn't be great for cross-subject analysis. What threshold to use
is a bit of an arbitrary decision, but the main idea is to remove spurious
voxels, so try different values until you're satisfied that it's doing a
decent job.

Overall, the general approach seems fine otherwise. The output of mrstats's
count value is the number of voxels in the mask you supplied. The numbers
in the file produced with the -dump option are just the voxel intensities
in the image within the mask. You supplied a 4D, 3 volume image, but you
get all the values for all 3 volumes as one big vector. You'd be better off
using mrstats for each of the FA, MD, AD & RD maps independently if you
need to use the -dump option...

Cheers,
Donald
Dear all,
I have a couple of questions regarding CSD tractgraphy and fiber
statistics. As I am new to this field, what I say is not clear and/or
doesn't make sense. If this is the case, my apolgy. Below you will see my
quesitons (Q) in the context.
I am trying to obtain MD, AD, and RD for the tracts we contrcuted using
CSD. I am aware that this question was raised before,
http://www.nitrc.org/pipermail/mrtrix-discussion/2014-May/000993.html, but
I would like to know the exact steps including command and paramters I
should be using. (Q1) I am also guessing that I need to obtain L1, L2, and
L3 and then calculate MD, AD, and RD in excel or something.  I would
apprecite if someone could help me out.
Here is the commands I think it's right at this stage. The CSD I ran was to
see the tracts between M1s. Note that I follow the steps described in here,
http://www.brain.org.au/software/mrtrix/tractography/tracking.html, which
is mainly to obtain FA. As such, some of the options may not make sense.

tensor_metric dt.mif -num 1,2,3 -value MD_val.mif
erode mask.mif -npass 3 - | mrmult MD_val.mif - - | threshold - -abs 0.4
sf_MD_val.mif
estimate_response dwi.mif sf_MD_val.mif -grad b.txt response_md.txt
csdeconv dwi.mif response_md.txt -lmax 8 -grad b.txt -mask mask.mif
CSD8_MD.mif
streamtrack SD_PROB CSD8_MD.mif -seed leftM1.mif -include ccmask.mif
-include rightM1.mif -mask mask.mif lm1_rm1_PROB_MD.tck
tracks2prob -template MD_val.mif lm1_rm1_PROB_MD.tck
FA_masklm1_rm1_PROB_MD.mif
threshold FA_masklm1_rm1_PROB_MD.mif FA_masklm1_rm1_PROB_MDthr.mif
mrstats MD_val.mif -mask FA_masklm1_rm1_PROB_MDthr.mif -dump
FA_masklm1_rm1_PROB_MDthr.txt
(Q2) Are these steps appropriate?
If I ran these commands, the command promt displays mean, sd, min, max,
count (Q3, what this count? fiber counts?). in three lines for ch0-2.  But
the text file contains only one columns. (Q4) What are these numbers in the
text file?
Thank you for your help in advance.
Matt

_______________________________________________
Mrtrix-discussion mailing list
Mrtrix-discussion at www.nitrc.org
http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20150723/a84a4f2c/attachment.html>