Graph analysis with the matrices obtained from fMRI data

[Shrikanth Kulashekhar]

Dear Braph Community,

My question if about using graph analysis on fMRI data. I performed an effective connectivity analysis using DCM and have connectivity configurations in terms of A,B and C matrices. I am wondering if I can now use a graph analysis on the DCM matrices to obtain graph metrics of the node interactions and a formal graphical representation of network.
If so, how should I proceed with this using Braph2.

Thank you,
Shrikanth

[Mite Mijalkov]

Hi Shrikanth,

Thank you for showing an interest in Braph 2.0 and hope that it will prove to be a useful tool for your research.

The answer to your question is yes, graph analysis can be used to analyze the DCM matrices and provide explanation of the topological properties of the nodes. You can do it with Braph 2. While the pipelines for such analyses are not implemented to the core release of the software, it comes with all the tools needed to perform this analysis (directed measures are implemented).

Together with the release of the software, We will post detailed explanations of how to implement additional pipelines in the software (it will be only several lines of code written in Matlab according to predetermined rules). On other hand, we can also assist you in implementing this pipeline.

If you have any additional questions in meantime, please do not hesitate to ask and we will try to help as much as possible.

Best,
Mite

[Shrikanth Kulashekhar]

Hi Mite,

Thank you for the information and help.

Since I am short on time to perform the graph analysis (for a revision), I can implement the pipleine for the analysis, with your help.

I downloaded the latest version of Braph 2 from the repository. From its folders I can see that the code for graph metrics and analysis is already there.

To begin, how can I proceed with moving the DCM data (from SPM) into Braph 2?

Shrikanth

[Mite Mijalkov]

Hi Shrikanth,

Of course, we can try to help you with the implementation of this pipeline.

To transfer the data from SPM software to Braph2, you would first need to export the data from SPM in .txt, .xls or .mat format. Then, if you have one .txt file per subject (that holds the connectivity matrix that is associated with the subject), you can group all the subjects from same group within a folder and import that folder in the cohort part of Braph2.

In order for us to be able to provide you with detailed information about how to proceed, we would need more information about the nature of the analysis that you want to perform. For example, given the A matrix from DCM for each subject, do you want to analyze the topology of this matrix? Alternatively, do you want to use also B and C matrices in some way?

I am looking forward to your response and will get back to you with more detailed explanation of how to proceed with the analysis.

Best,
Mite

[Shrikanth Kulashekhar]

Hi Mite,

Thats great! Thank you.

We want to analyse the graphs and its metrics using the full A-B-C connectivity matrices.

I have a basic question about using the matrices for each subject, would it be possible to just use the (single) group level A-B-C matrices for the graph analysis?

The reason I ask is because in our DCM analysis we did not have a clear hypothesis about the network architecture for the five task relevant brain regions. We therefore used a “data-driven” approach, with Parametric Empirical Bayesian method (PEB) to identify the connectivity structure. As far as I understand, PEB seeks the optimal model by testing various combinations of A-B-C-parameters through a pruning process, the winning model is therefore estimated at the group level instead of the individual subjects.

Let me know if you need any clarifications or more information.

Shrikanth

[Shrikanth Kulashekhar]

Hi Mite,

I was wondering if you had time to think about my query, on how to proceed from the group DCM to graphs. If the estimated effective connectivity matrix from DCM can be interpreted as a weighted adjacency matrix of a directed graph between the brain regions. For the Braph pipeline implementation, the group DCM could be treated as a single subject, would this be feasible?

Looking forward to your feedback.

Thank you,
Shrikanth

[Mite Mijalkov]

Hi Shrikanth,

Thank you for sharing the information regarding your analysis. I have met with my colleagues and we thing that this analysis can be done in Braph.

In order to know how to proceed with the implementation we would need several technical details regarding your analysis. I was wondering whether you would be available for a short meeting so that we can discuss them and decide on a plan on how to proceed with the implementation? Please contact me on my e-mail so that we can agree on a meeting time.

Looking forward to your response and the meeting.
Best,
Mite

[Shrikanth Kulashekhar]

Hi Mite,

Thank you for taking the extra effort to help with the analysis. I have sent you an email.

Shrikanth

[Yu-Wei Chang]

Hi Shrikanth,

I hope the emails help a bit 🙂
If applicable, please follow us @braph2software on twitter and cite us (Braph & Braph2 GitHub)

Best,
Yu-Wei

[Author]

Posts