Electroencephalography (EEG) is a primary tool for diagnostics and follow-up of neurological conditions. From EEG data, it is possible to obtain functional connectivity matrices, enabling the identification of associated brain areas during a task, hence supporting hypothesis for specific brain links. However, usually these matrices are built using correlation tools, which are not appropriate when characterizing nonlinear signals such as EEG. Hence, here we apply the same procedure using information theoretical measures (mutual information, transfer entropy and causal mutual information) and compare the results. For the test case applied, we show that both ways lead to similar links identified, however the information-theoretical measures provide the extra indicator of the direction to which the information flows.
Noslen Hernández, Raymundo Machado de Azevedo Neto, Aline Duarte, Guilherme Ost, Ricardo Fraiman, Antonio Galves, Claudia D. Vargas
Using a new probabilistic approach we model the relationship between sequences of auditory stimuli generated by stochastic chains and the electroencephalographic (EEG) data acquired while participants are exposed to those stimuli. Herein, the structure of the chain generating the stimuli is characterized by a rooted and labeled tree whose branches, henceforth called contexts, represent the sequences of past stimuli governing the choice of the next stimulus. A classical conjecture claims that the brain assigns probabilistic models to samples of stimuli. If this is true, then the context tree generating the sequence of stimuli should be encoded in the brain activity. Using an innovative statistical procedure we show that this context tree can effectively be extracted from the EEG data, thus giving support to the classical conjecture.
Nastaran Lotfi, Antonio J. Fontenele, Thaís Feliciano, Leandro A. A. Aguiar, Nivaldo A. P. de Vasconcelos, Carina Soares-Cunha, Bárbara Coimbra, Ana João Rodrigues, Nuno Sousa, Mauro Copelli, Pedro V. Carelli
It has recently been reported that statistical signatures of brain criticality, obtained from distributions of neuronal avalanches, can depend on the cortical state. We revisit these claims with a completely different and independent approach, employing a maximum entropy model to test whether signatures of criticality appear in urethane-anesthetized rats. To account for the spontaneous variation of cortical state, we parse the time series and perform the maximum entropy analysis as a function of the variability of the population spiking activity. To compare data sets with different number of neurons, we define a normalized distance to criticality that takes into account the peak and width of the specific heat curve. We found an universal collapse of the normalized distance to criticality dependence on the cortical state on an animal by animal basis. This indicates a universal dynamics and a critical point at an intermediate value of spiking variability.
Cecilia Romaro, Antonio Carlos Roque, Jose Roberto Castilho Piqueira
There is a strong nexus between the network size and the computational resources available, which may impede a neuroscience study. In the meantime, rescaling the network while maintaining its behavior is not a trivial mission. Additionally, modeling patterns of connections under topographic organization presents an extra challenge: to solve the network boundaries or mingled with an unwished behavior. This behavior, for example, could be an inset oscillation due to the torus solution; or a blend with/of unbalanced neurons due to a lack (or overdose) of connections. We detail the network rescaling method able to sustain behavior statistical utilized in Romaro et al. (2018) and present a boundary solution method based on the previous statistics recoup idea.
Iara Frondana, Rodrigo R.S. Carvalho, Florencia Leonardi
Epilepsy has been a central topic in computational neuroscience, and in silico models have shown to be excellent tools to integrate and evaluate findings from animal and clinical settings. Among the different languages and tools for computational modeling development, NEURON stands out as one of the most used and mature neurosimulators. However, despite the vast quantity of models developed with NEURON, a fragmentation problem is evident in the great majority of models related to the same type of cell or cell properties. This fragmentation causes a lack of interoperability between the models because of differences in parameters.
Featuring this week:
Stay informed on our latest news!
|Follow Us on Facebook|