Print

Print
Hi Sim Luck,

Please see below:


On 14 February 2017 at 17:52, Sim Luck <[log in to unmask]> wrote:
Hi, Anderson,

I just found your message - thank you very much.

Regarding the example "to obtain it as a probabilistic map", I do not find where you used "10" in the command: fslroi, ${FSLDIR}/data/atlases/HarvardOxford/HarvardOxford-sub-prob-?mm.nii.gz. Then where should I define this 10?

Thy something as this:

fslroi ${FSLDIR}/data/atlases/HarvardOxford/HarvardOxford-sub-prob-2mm.nii.gz mymask.nii.gz 9 1
 

Further, is that -thr 10 only for the HarvardOxford-Subcortical.xml in ${FSLDIR}/data/atlases? Can I use this to any atlases? Could you give me a full example for this fslmaths application (using, -thr, -uthr, and -bin)?

All atlases are stored in a similar way, with 3D, 4D, and .xml files. 

Try opening (with FSLview, FSLeyes, Mricron, Mango, etc) the image files that contain these atlases. You'll see that some files are 4D files with probabilities (one region per volume), whereas others are crisp, with probabilities set at some levels (e.g., 50%), with each region identified by a unique integer in a single 3D volume.

For the files that are 4D, you'd use fslroi. For the files that are 3D you'd use fslmaths. I think all atlases are represented in both ways.

All the best,

Anderson

 

Thank you very much,
Sim Luck



On Wed, Feb 8, 2017 at 4:20 AM, Anderson M. Winkler <[log in to unmask]> wrote:
Hi Sim Luck,

In the HO atlas, each region has an index. See the file HarvardOxford-Subcortical.xml in ${FSLDIR}/data/atlases.

Amygdala is #9, but counting starts at zero, so it's the 10th region.

To obtain it as a probabilistic map, one would take the 10th volume (along the 4th dimension) of the file ${FSLDIR}/data/atlases/HarvardOxford/HarvardOxford-sub-prob-?mm.nii.gz using fslroi.

To obtain as a binary map, one approach is to extract just the voxels with value equal to 10 from the file HarvardOxford-sub-maxprob-thr*-?mm.nii.gz, and make a binary mask. This can be done with the options -thr and -uthr, and then -bin from fslmaths.

Hope this helps.

All the best,

Anderson


On 7 February 2017 at 20:12, Sim Luck <[log in to unmask]> wrote:
Hi, Anderson,

Any advice for me on the Threshold issue with ROI anlaysis?

Thank you very much,
Sim Luck

---------- Forwarded message ----------
From: Sim Luck <[log in to unmask]>
Date: Mon, Feb 6, 2017 at 2:19 PM
Subject: How to use -thr in ROI mask
To: FSL - FMRIB's Software Library <[log in to unmask]>


Hi, FSL experts,

I am learning to create the ROI, using FSLview. For instance, based on Harvard-Oxford Subcortical Structural Atlas, I got the ROI: Amy.nii.gz. Definitely, this is based on anatomy. My question is whether I should use the command below to threshold it.

fslmaths  Amy.nii.gz  -thr 10  -bin   Amy.nii.gz_Thr.nii.gz

If so, how to decide the threshold? Why it should be 10? Or does it mean that people use this command as they create the ROI mask based on their contrast (e.g, a t-test)? I assume that this kind of ROIs are based on the activated coordinates. Any comments are much appreciated.

Thank you very much,
Sim Luck