Difference between revisions of "Resting state brain activity system identification framework (RestSYSID)"

Latest revision as of 09:12, 9 September 2013

[edit] People

• Javier Guaje
• Francisco Gómez

[edit] Summary

Resting state provides a flexible protocol to investigate cerebral activity. The study of this dynamic has provided relevant information for several pathological/pharmacological brain conditions such as, Alzheimer, Parkinson and disorders of consciousness and anesthesia, among others. Despite of their extended use of this experimental protocol, the underlying bases that rules this large brain scale coordinated activity remain poorly understood. In the last years, some models based of the synchronization of neuronal mass units have been proposed to explain this emergent phenomenon. Nevertheless, the parameterization of these models and the data manipulation need to reproduce empirical observations are quite complex. In addition, the exact mechanism that supports or suppresses coordinated activity is not clear in this kind of models. These limitations difficult considerably their to formally study altered brain conditions. Recently, a series of models based on mechanical statistics have been proposed to explain the emergent phenomena behind resting state activity. The main idea is to use mechanical statistical models with very simple interactions, for instance the Ising models, to reproduce the emergent coordinated activity, typically observed in resting activity. The main advantage of this modeling strategy is their simplicity for the parameterization and the computational implementation. Additionally, these models operate on binary state representations of the brain activity, which simplify considerably the data treatment to reproduce experimental observations. In other hand, the study of brain activity has been greatly influenced by system identification approaches, such as dynamic causal modeling. This approach provides a standard framework to compare brain dynamic hypotheses linking brain models parameters and experimental data. Unfortunately, these system identification approaches have not been implemented yet for resting state activity, mainly, because of the computational problem of exploring milliards of dynamics based on classical neuronal mass models. In this work, we will propose a novel framework to study brain activity based on resting state mechanical statistical models. The framework is going to allow 1) system identification, 2) model comparison and 3) family model comparison on empirical data and is going be tested in the scenario of anesthesia mechanism understanding.

[edit] Method overview

[edit] Data sources

[edit] Results (Expected)

• Method conference. Framework proposal.
• Journal article. Complete framework report.

[edit] References

Changes in Effective Connectivity by Propofol Sedation. F. Gómez*, C. Phillips*, A. Soddu, M. Boly, P. Boveroux, A. Vanhaudenhuyse, M-A Bruno, O. Gosseries, V Bonhomme, S. Laureys, Q. Noirhomme. PLOS One. (2013).

Ising-like dynamics in large-scale functional brain networks. Phys Rev E Stat Nonlin Soft Matter Phys. 2009 June; 79(6 Pt 1): 061922.

http://www.scholarpedia.org/article/Dynamic_causal_modeling