Mohammad Joudy: Inference of connectivity from recordings of brain activity
| When |
May 11, 2021
from 05:15 PM to 05:45 PM |
|---|---|
| Where | Zoom Lecture. Meeting ID and password will be sent with the invitation. You can also ask Fiona Siegfried for the access data. |
| Contact Name | Fiona Siegfried |
| Contact Phone | 0761 203 9549 |
| Add event to calendar |
 vCal iCal
|
Abstract
The connectivity of networks formed by neurons and synapses is key to understand function and dysfunction of the brain. The concept of “effective connectivity” addresses the causal interactions for a given set of neuronal activity signals recorded from the brain. Inference of causation from correlation is problematic, however, as different networks can yield the same correlations. To disambiguate the networks inferred from fMRI data an additional sparsity prior has been imposed. This method, however, does not perform well for “fast” signals obtained with ECoG or Calcium imaging. Fitting a linear dynamics (or an auto-regressive process), in contrast, can work well in such cases. In order to validate and assess the performance of such a method, we simulated a large network of spiking neurons (“Brunel network”) in the asynchronous-irregular regime to provide some ground truth data. As a surrogate for calcium imaging data, we lowpass filtered the neuronal spike trains. We show that methods based on temporal precedence can achieve a better estimation of underlying network. As a result, depending on the type of data to be analyzed, different methods of inference should be applied.