open-discussion > Network matrix values different
Showing 1-4 of 4 posts
Results per page:
Mar 7, 2017  07:03 AM | Nabin Koirala - JGU
Network matrix values different
Dear PANDA team , 

I used PANDA toolbox to compute the probabilistic tractography network matrix using AAL (116 ROIs) atlas, and I used the FSL independently to get the same matrix and somehow the values in the network matrix for both are quiet different. The matrix obtained from PANDA has almost 10 times higher values than I get when I do it only using FSL.So, I was wondering if PANDA is overestimating something or I am making some silly mistake in my processing in FSL ? 

This is the processing I did in FSL. 

1. run eddy current then bet
2. DTIfit
3. Bedpostx
4. Probtrackx2 command with --network option and --onewaycondition and steplength 0.5, 5000 streamlines
5. The obtained fdt_network matrix is then divided with waytotal matrix rowwise to obtain the final matrix. 

Mar 9, 2017  12:03 AM | Zaixu Cui
RE: Network matrix values different
Dear Nabin,

It may be caused by the data processing procedure difference.

For example, after eddy current, you should rotate the b vectors according to the affine matrix generated. 


Mar 9, 2017  01:03 AM | Nabin Koirala - JGU
RE: Network matrix values different
Dear Zaixu,

Thank you for the reply. So would it be possible to point me to the right direction here by mentioning the steps that PANDA does (atleast in brief as you mentioned above) or you could also tell me the matlab scripts / or other documentation where I should look and could find these details. It would be really helpful and greater assurance to continue using PANDA.

Many thanks again in advance.


Mar 9, 2017  08:03 PM | Zaixu Cui
RE: Network matrix values different
Dear Nabin,

Thank you for your interest in PANDA.

We have elaborated the data processing procedure in our paper:

From the original data to network construction, there are many procedures. Also, the brain network node definition is important for the final results.

All the MATLAB scripts were named as 'g_*.m' in the folder of PANDA. For example, about the brain network node definition, the function is g_IndividualParcellated.m.
About the full pipeline, you can find the g_dti_pipeline.m function. In this script, you can find the order of the full data procedures.

If you have any problems, feel free to contact me.

Best wishes