want information regarding brain tumor dataste in TVB

[Maria Nazir]

  Dear TVB users

 In the TVB brain tumor dataset, the authors have provided structural connectivity of 11 normal healthy control subjects before as well as after surgery .. but its not making sense as health controls did not go through surgery then why they took their structural connectivity again instead of considering the same pre-surgery SC matrix  of  normal subjects??  I am confused.

Secondly, can i compare the functional connectivity of glioma patients after surgery with the average FC of the control subjects  ??? is it legit ? or I read somewhere that FC is variable across subjects so it can't be compared with different subjects ?? Is that true ??

Thirdly, is there any authentic metric for comparison of structural connectivity between healthy and control subjects ?? I am not talking about pearson's correlation as it gives us a matrix . I want some scalar number that can give us a clue about how much these two matrices are closer or far apart. 

Maria Nazir

TEAM LEAD

Medical Imaging and Diagnostics Lab 

National Center of Artificial Intelligence  

COMSATS ISLAMABAD

[Daniele Marinazzo]

Dear Maria

control subjects (I would not call them "normal" for obvious reasons) underwent two scans, corresponding to the pre- and post- surgery for the tumor patients.

The other two questions depend on the scientific question you have in mind.

The way TVB compares empirical and simulated FC matrices (which can be used for other matrices too of course), is to vectorize the upper or lower triangular part (for asymmetric matrices both, removing the diagonal if it's the same for all matrices), and then Pearson correlation between these vectors. Many other choices are possible.

--
You received this message because you are subscribed to the Google Groups "TVB Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tvb-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAAhQMeUqC3UPY54u23km5eGx_Efr_4t4Pjhz1S9NrLgs1%2BK%2BAg%40mail.gmail.com.

[Maria Nazir]

Dear TVzb users, 

I want to know about the TVB Brain tumor dataset that has been made public for pre and post surgery patients of brain tumor. The derivatives folder (that i believe came after pre-processing raw MRI, FMRI and DWI ) for both pre nd post suergery patients contains the files as mentioned in the image below. 

I have many queries related to this data. Please explain one by one for better understanding.

1. Can anyone  working with this data , explains to me the difference between SCthrAn.mat and SC.zip. ???

2. secondly I am unable to load the FC.mat file in TVB. how can i do that  exactly ??

3. Also What is the difference between this FC.mat file ( as shown in the image above ) and the FC dynamic matrix we get from first inserting SC.zip , simulating using reduced wong wang and then finally simulating bold monitor in TVB desktop application using the pearson correlation coefficient. ??? Are they the same ?

4.  What is the use of this HRF

5. If we want to simulate brain network model, is SC.zip enough ???

Thanks in advance for the help.

To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAH%2BK%3D0ina1don5x_PT8OzGjQUYiJ_9T-oN%2BCU%2BrVJ6ZFjV4L3w%40mail.gmail.com.

[Daniele Marinazzo]

Hi

some replies inline

 

I have many queries related to this data. Please explain one by one for better understanding.

1. Can anyone  working with this data , explains to me the difference between SCthrAn.mat and SC.zip. ???

The zip folder contains more information, see the data descriptor (also attached to this email for convenience)

 

2. secondly I am unable to load the FC.mat file in TVB. how can i do that  exactly ??

You don't. That's the empirical FC.

 

3. Also What is the difference between this FC.mat file ( as shown in the image above ) and the FC dynamic matrix we get from first inserting SC.zip , simulating using reduced wong wang and then finally simulating bold monitor in TVB desktop application using the pearson correlation coefficient. ??? Are they the same ?

No, the one you get from TVB is the modelled FC. The other is the empirical one.

 

4.  What is the use of this HRF

The idea is that a region- and subject-specific HRF, as the one that one can derive from the data using the approach described here

https://pubmed.ncbi.nlm.nih.gov/34560269/

can provide a more precise fit to empirical data. Some tests described below

https://github.com/AmoghJohri/TVB-Tests/blob/master/legacy_vs_rsHRF%20_filtered.ipynb

 

5. If we want to simulate brain network model, is SC.zip enough ???

yes. Then if you want to fit it to the empirical FC you need the FC.mat too. Otherwise you can just maximize metastability/excitability without comparing to the empirical FC. 

Thanks in advance for the help.

cheers 

To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAAhQMeUWMq%2BK3gc%2BO2zeSwOtvJgdVHg-0Rup5jv1hvzVEk0ebA%40mail.gmail.com.

[Maria Nazir]

Dear Daniele,

Thankyou so very much for such a detailed reply. Will be pokng you again for more help.

Maria Nazir

TEAM LEAD

Medical Imaging and Diagnostics Lab 

National Center of Artificial Intelligence  

COMSATS ISLAMABAD

To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAH%2BK%3D0iPGRaefU3WgOYq4dTD%2BZAyO1%2BP-oFGmv6tV2jHO8fRDA%40mail.gmail.com.

[Maria Nazir]

Dear Daneile,

Is the TVB preprocessed  data ( SC.zip, SCthrAn and FC.mat ) enough  for simulations and modelling in TVB ( If we just do some comparisons by simulating each subjects SC.zip using different neural mass models i.e, jansen ritt or reduced wong wang etc) ???? or do we have to again do the preprocessing of raw TVB data using tvb pre-processing pipeline and again extract SC, FC and bold ??

Maria Nazir

TEAM LEAD

Medical Imaging and Diagnostics Lab 

National Center of Artificial Intelligence  

COMSATS ISLAMABAD

On Wed, 8 Feb 2023 at 20:55, Daniele Marinazzo <daniele....@gmail.com> wrote:

To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAH%2BK%3D0iPGRaefU3WgOYq4dTD%2BZAyO1%2BP-oFGmv6tV2jHO8fRDA%40mail.gmail.com.

[Daniele Marinazzo]

enough

To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAAhQMeXtHJe6kXtsb4ZYUveU%2BUU88fFPoTRrew%2BiMNoymrNy3w%40mail.gmail.com.

[Maria Nazir]

Dear Daniele ,

I hope you are doing well. I did some simulations and want to ask some questions from you related to the image shown below.  The image show some simulation results on the right.

So I simulated SC.zip and got the time series bold by simulating the bold monitor. Then from the analysis tab, I went to the functional connectivity dynamic matrix calculator and got two FCD as can be seen in the above image so i am confused which one is the simulated FCD matrix that can be compared with the empirical FCD already given with the brain tumor TVB data ?? 

Secondly, I also found pearson correlation coefficients using the analyzer tab of the time series bold which I originally got as i told above by simulating SC. zip. The correlation coefficients data type can also be seen in the figure above. FC is actually found by calculating the Pearson correlation coefficients . So are these correlation coefficients the required simulated FC that I am finding or the previous one ?? I am literally confused about these two ( 2 FCD matrix and Pearson correlation coefficients ). If they both are the same thing then why is the TVB web interface giving us both options and which one is the right FC matrix  to use for further analysis and comparison with the empirical FC.mat ??

Your help is highly needed and appreciated.

Maria Nazir

TEAM LEAD

Medical Imaging and Diagnostics Lab 

National Center of Artificial Intelligence  

COMSATS ISLAMABAD

To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAH%2BK%3D0hjArTH-Qjx7rpPPNBd7V-0g8XjsNOoWrZMVb6VhVMm4g%40mail.gmail.com.

[Daniele Marinazzo]

Dear Maria

I am not an expert on the web interface, I'll leave the answer to others.

Cheers

To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/CAAhQMeUboqi6RCP%2BxfxDKwtMC61EZxYEbntatNSVawx-3HXTxg%40mail.gmail.com.

[Lia Domide]

Dear Maria,

All of the results from your screenshot (TimeSeriesBold, 2 FCD, PearsonCorrCoef) are simulated.

They were computed with TVB code. You can read about those algorithms in our public repo:

Pearson Correlation Coef: 

https://github.com/the-virtual-brain/tvb-root/blob/master/tvb_framework/tvb/adapters/analyzers/cross_correlation_adapter.py#L379

https://numpy.org/doc/stable/reference/generated/numpy.corrcoef.html

FCD:

https://github.com/the-virtual-brain/tvb-root/blob/master/tvb_framework/tvb/adapters/analyzers/fcd_adapter.py#L111

Best,

Lia.