Down-sampling from 30K cortex to 2K

Bugs
1

Dear BS developers,

Today I noticed that (latest version of) Brainstorm destroys the structure of cortex during down-sampling.
For instance I wanted to down-sample the 30K ICBM cortex to 2K. below you can find a link for the plot.

I would appreciate if you let me know how to fix this problem.
Best,
Keyvan

2

While Francois is on a deserved break, I’ll try answering. Did you use “remesh” or “less vertices”? I’m running the Aug 28 version and I have no problems with less vertices, from 15,000 vertices to 2,000 vertices. If I use “remesh”, it correctly works, but is not very useful, since it is not intended for the undulating cortical surface.

3

Hi,

I am using the version currently available on the website. below is the link:
http://neuroimage.usc.edu/bst/download.php
Please use “less vertices” and down-sample from 30/15K to 2K, then plot it. However, I fixed the problem for my case using older version of BS.

Best wishes

4

Hello,

I changed a few things in the way the surfaces were downsampled, but I cannot reproduce the error you reported.
I would like to understand where this problem is coming from rather than recommending you to use an older version of Brainstorm.

Can you please do the following:

  1. Make sure you are running the latest version of Brainstorm (today)
  2. Make sure you do not have any other version of Brainstorm in your Matlab path (if the brainstorm folders are in your saved path, delete them all).
  3. Create a new protocol
  4. In the default anatomy, right-click on cortex_15000V > Less vertices > 2000 vertices
  5. Check the output surface

If you still have this problem, could you send me this protocol? It might help me understand what the problem is.
Click on the menu File > Export protocol > Export as zip file. Then upload it somewhere (eg. dropbox) and send it to me in a separate email (click on my username on this forum).

Can you also copy-paste here the output of the command “ver” in your Matlab command window?

Thanks,
Francois

5

Hi Francois,

Yes, I use the firs option (reducepatch) for down-sampling.
Below is the output of ver command in Matlab:

>> ver

MATLAB Version: 8.5.0.197613 (R2015a)
MATLAB License Number: 590253
Operating System: Linux 3.13.0-24-generic #47-Ubuntu SMP Fri May 2 23:30:00 UTC 2014 x86_64
Java Version: Java 1.7.0_60-b19 with Oracle Corporation Java HotSpot™ 64-Bit Server VM mixed mode

MATLAB Version 8.5 (R2015a)
Bioinformatics Toolbox Version 4.5.1 (R2015a)
Curve Fitting Toolbox Version 3.5.1 (R2015a)
Image Processing Toolbox Version 9.2 (R2015a)
MATLAB Compiler Version 6.0 (R2015a)
MATLAB Compiler SDK Version 6.0 (R2015a)
Optimization Toolbox Version 7.2 (R2015a)
Signal Processing Toolbox Version 7.0 (R2015a)
Statistics and Machine Learning Toolbox Version 10.0 (R2015a)
Symbolic Math Toolbox Version 6.2 (R2015a)

Keyvan

6

Hi Keyvan,

I realized that one of the modification I made was only working with Matlab 2016a… I think I fix this issue.
Please update Brainstorm and try again.

Francois

7

Hi Francois,

That’s OK now.

Thanks a lot,
Keyvan

8

Interesting. I had no problems under 2015b.

9

Ok, I guess it was working with 2015b and 2016a, and it was doing weird things before 2015a.
(the interesect function now returns the indices in a different order, the outputs are sorted by default)