Workshop: Spatiotemporal Dynamics in Neuroimaging: Models and Analysis - Q & A

This space is to organise knowledge (Q&A) around the topic of models and analysis of spatiotemporal dynamics that will be central theme of our workshop at CNS*2020 Online.

Feel free to start sending questions, we’re thinking about discussing them at the end of the workshop :nerd_face:


Session 01 on neural-flows toolbox
@John_Griffiths asked: Would it be interesting and/or useful to use the toolbox to analyze heart neural/muscle activity? Exquisite 3D waves in cardiac tissue.

The cardiology people are ahead of us! Lot of inspiration for the toolbox came from the field of computational cardiology. Indeed exquisite waves in the heart :heartbeat:

Session 01 on neural-flows toolbox

@James_Henderson asked Are there any assumptions that this will be used for brain data, or can it be used for any system?

1 Like

Session 02 by Changsong Zhou on Hierarchical Connectome Modes and Critical State Jointly Maximize Human Brain Functional Diversity

@Shrey_Dutta asked: That exponent of power law (e.g. 1.5) makes the system critical and why? What’s the rational behind the chosen/observed exponent (magical number) other than that it comes from the universality classes?

@James_Henderson: the framework itself can be applied to any system – provided that the intensity/amplitude is smooth – and its first and second spatiotemporal derivatives are also smooth.

The toolbox itself sort of expects brain data. There is a step in which we estimate the brain’s convex hull enclosing the scattered measurement points (sources, or centres of gravity in parcellated data). This is essentially a boundary which is used in the flow estimation. I’m not sure the boundary estimation would work very well for more complex 3D shapes for instance.

Session 03 by @yujiangwang Spatio-temporal dynamics in epilepsy: Models & Analysis
The data and code used in the work she presented can be found here:

Q: Did you try to look at the (intrinsic) dimensionality of the seizure space?

Yes, using stability NMF we obtained the optimal number of dimensions, which was slightly different between all patients. Details are in the paper: