Computational Modelling Group

• Login/Register

You are here:

• CMG Home
• Events
• Dynamic Causal Modelling for M/EEG

Web page

https://iris.ucl.ac.uk/research/personal?upi=LITVA78

Submitter

Luke Goater

Dear all,

You are kindly invited to a free seminar at the Institute of Sound and Vibration Research (ISVR) on “Dynamic Causal Modelling for M/EEG”presented by Dr. Vladimir Litvak a Senior Research Associate at the Welcome Trust Centre for Neuroimaging from UCL Institute of Neurology. Dr. Litvak is responsible for developing EEG/MEG analysis methods and support of the SPM software package at UCL Institute of Neurology. More details can be found here

As we are trying to have an estimate of the number of people attending, if you are interested, please send an email to pth1v10@soton.ac.uk I would appreciate it if you would circulate this email to the people you think are interested in this topic.

Venue: Room 3019, Institute of Sound and Vibration Research (ISVR), University of Southampton

Date: Tuesday 08 May 2012

Time: 12:00-13:00

Dynamic Causal Modelling (DCM) is a framework bringing together data analysis and neural modelling. In DCM measured data are explained by a network model consisting of a few sources, which are dynamically coupled. This network model is fitted to the data using a Bayesian inversion scheme. Practically this means that what is being optimized is not only the goodness of data fit but also the deviation of model parameters from their expected prior values (e.g. physiologically meaningful range). Model inversion provides two main results. The model evidence is a single number specific to the combination of data and model which can be used to compare models and test specific hypotheses. The posterior density on model parameters can be used to make inferences about connections between sources or their condition-specific modulation under the model selected. For M/EEG data, DCM can be a powerful technique for inferring (neuronal) parameters not observable with M/EEG directly. Specifically, one is not limited to questions about source strength, but can test hypotheses about connections between sources in a network. In the recent years, several variants of DCM for M/EEG have been developed for modelling evoked responses, steady state power and cross-spectra, induced responses and phase coupling.