The paper that you cite makes me ask: are the JHU atlases used for tractography and for white matter labels obtained through the FSL distribution in the correct R-L orientation and labeled appropriately?

Stefano Marenco, MD
NIMH/CBDB
10 Center Drive, Bldg 10 room 3C103
Bethesda MD 20892
Tel 301 435-8964
Fax 301 480-7795
Email: [log in to unmask]


-----Original Message-----
From: Yaroslav Halchenko [mailto:[log in to unmask]]
Sent: Thursday, January 10, 2013 12:55 PM
To: [log in to unmask]
Subject: Re: [FSL] Incorrect probabilities in Harvard-Oxford-sub Left hemisphere

NB a regular friendly monthly buzzzzz

is anyone interested to do bits more of reverse-engineering
analysis and co-author a little paper similar to

http://www.frontiersin.org/Brain_Imaging_Methods/10.3389/fnins.2013.00004/full
Torsten Rohlfing
Incorrect ICBM-DTI-81 Atlas Orientation and White Matter Labels

;-)


On Tue, 04 Dec 2012, Yaroslav Halchenko wrote:

and just to keep things at least somewhat luke warm (thus cooking), here

is another visualization summarizing the problem.

I took left hemisphere, mirrored right hemisphere (so more or less now they

should be 'corresponding'), resorted areas from what they were in atlas

so they are in correspondence as well

for each voxel in each of the two hemispheres:

- computed 'max probability' area for each voxel and assigned index

  from

new index    old index    area

0,              7,  'Brain-Stem')),

1,              10, 'Left Accumbens')),

2,              9,  'Left Amygdala')),

3,              4,  'Left Caudate')),

4,              1,  'Left Cerebral Cortex ')),

5,              0,  'Left Cerebral White Matter')),

6,              8,  'Left Hippocampus')),

7,              2,  'Left Lateral Ventrical')),

8,              6,  'Left Pallidum')),

9,              5,  'Left Putamen')),

10,             3,  'Left Thalamus'))

- added %probability/100 to each index, so e.g. value of

  4.67 is "Cerebral Cortex" at 67%

scatter plot left hemisphere against right flipped:

http://www.onerussian.com/tmp/test-left+right_rev.png

you can easily see that while some areas seems  to be Ok (e.g. brainstem ;) )

others "leak" too heavily among each other, and "right" probabilities

tend to be generally higher than left for each one of those areas

On Wed, 07 Nov 2012, Yaroslav Halchenko wrote:

On Wed, 31 Oct 2012, Yaroslav Halchenko wrote:

On Wed, 31 Oct 2012, Stephen Smith wrote:

  Hi Yaro - thanks for letting us know.  We'll look into this ASAP and

  update this if necessary.

Great -- thanks in advance!

And while you are at it, might be worth fixing a spelling typo in the

<type> of those:

/usr/share/fsl/data/atlases/Cerebellum_MNIflirt.xml:    <type>Probabalistic</type>

/usr/share/fsl/data/atlases/Cerebellum_MNIfnirt.xml:    <type>Probabalistic</type>

/usr/share/fsl/data/atlases/HarvardOxford-Cortical.xml:    <type>Probabalistic</type>

/usr/share/fsl/data/atlases/HarvardOxford-Subcortical.xml:    <type>Probabalistic</type>

/usr/share/fsl/data/atlases/JHU-tracts.xml:    <type>Probabalistic</type>

/usr/share/fsl/data/atlases/Juelich.xml:    <type>Probabalistic</type>

/usr/share/fsl/data/atlases/MNI.xml:    <type>Probabalistic</type>

/usr/share/fsl/data/atlases/Thalamus.xml:    <type>Probabalistic</type>

--
Yaroslav O. Halchenko
Postdoctoral Fellow,   Department of Psychological and Brain Sciences
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik