EEG Analysis Sample Questions? Flashcards
In EEG, referencing is done by …….. the average activity at the reference electrodes from all other channels
subtracting
During preprocessing EEG, high pass filtering is done usually on ….data instead of epoched data to avoid edge artefacts..
continuous
Independent component analysis separates a set of signal mixtures into a corresponding set of statistically independent component signals of…… signals
source
In the equivalent current dipole approach for EEG source reconstruction, the locations and ……. of dipoles remain fixed over time, but their magnitude may vary
orientations
For an EEG signal to be considered chaotic at least one of its Lyapunov exponents should be….
positive
In non linear dynamical system theory, a strange attractor is a complex object with….. geometry
fractal
While using magnitude square coherence for analysing synchrony in the EEG data, the major problem is that it ……….. amplitude and phase information
mixes
in brain the effective connectivity describes the …… effects of one brain region over another
casual
A brain network with a ……. characteristic path length means that information, on average, be routed between distant brain regions using only a few intermediate regions
short/low
Choose the INCORRECT statement in relation with muscular artifact in EEG recording
a. it contaminates high frequency brain activity
b. problematic for time frequency analysis
c. less problematic for ERP analysis
d. it is mostly found over the midline electrodes
d.it is mostly found over the midline electrodes
Choose the INCORRECT statement in relation with global field power
a. relates to mean root square
b. it is dependant on the choice of EEG reference
c. it is visualised as a function of time
d. high values indicate a stable configuration of EEG scalp map
b.it is dependant on the choice of EEG reference
In relation to scaling exponent a. using detrended fluctuation analysis, choose the incorrect statement
a. a=0 indicates white noise
b. a=1 indicates 1/f noise
c. 1<a></a>
a. a=0 indicates white noise
In relation with EEG analysis, choose the odd one eout
a. correlation dimension
b. lyapunov eponent
c. entropy
d. fourier power spectrum
d. fourier power spectrum
Choose the incorrect statement in the context of Granger causality analysis for EEG signal
a. it provides directional information
b. it indicates mainly nonlinear coupling
c. the estimation is biased for limited data
d. the formulation is parametric
b. it indicates mainly nonlinear coupling
Choose the incorrect statement in the context of phase synchrony analysis for EEG signals
a. phase synchrony is a neurophysiologically meaningful analysis
b. it is sensitive to weak coupling
c. it requires fewer assumptions
d. phase is meaningful only for broadband EEG signals
d. phase is meaningful only for broadband EEG signals
Discuss why mean amplitude measure may be preferred over peak amplitude measure in ERP analysis.
The peak amplitude measure is very sensitive to high frequency noise and it is a nonlinear measure and therefore the average of the peak amplitude cannot be compared to the grand average. IN contrast the mean amplitude represents the data in a better way as it characterizes as being extended over time, as the effect may be nowhere near the peak. It is also linear and can be compared to the grand average and can use a narrower measurement window
What is a forward problem in the EEG source reconstruction? Discuss briefly the factors influencing the accuracy of the forward problem solution.
The forward problem is asking the question of what will be the source given the dipoles. It has a unique solution and can be solved analytically without EEG data. It accuracy depends on how is the head modeled. Therefore, the solution may differ based on whether the head is represented as anatomically correct or as a sphere. Also, it depends on whether data from each participants is measured precisely or standard electrode montage is assumed.
Discuss briefly the common strategies used for the Equivalent Current Dipole approach for EEG source reconstruction.
One way would be to use PCA to find out the underlying spatial distributions of the activity.
Another way is to start with a few sensory dipoles to fit the early ERPs and then add more for later time windows.
Also, preexisting strategies can be used to decide the number of dipoles
State the limitations of ECD approach for EEG source reconstruction.
It is not possible to know the priori number of dipoles
Most cortical activity is distributed over a large body of the cortex
The results are dependent on the researcher as there are many free parameters to choose and the result is dependent on these choices which can be biased by expectation
Cannot be completely sure about the accuracy of the solution as with so many free parameters per dipole, if one is wrong the rest will slightly change to reduce the residual variance. Therefore, if there is noise in the data, a wrong solution can have lower variance than the best fit solution
According to the dynamical system theory, what is an attractor? Name the four common attractors.
An attractor is a geometrical figure, around which a dynamical system evolves over long enough time. There are 4 types of common attractors:
Fixed point – the system cannot change anymore without external force
Limit cycle – a closed loop in the state space
Torus – has complex donut shape
Strange attractor – complex object with a fractal geometry, associated with chaotic dynamics
Name the two crucial parameters for state space reconstruction of an EEG signal. Name the methods used for selecting these parameters.
The two crucial parameters in reconstructing state space are time delay and embedding dimension. Some of the approaches that can be used to choose these parameters are:
Embedding – time series is converted into a sequence of vectors in an m-dimensional space. There are two types of embedding (time-delay and spatial embedding)
Correlation dimension, lyapunov exponent
Discuss briefly why brain can be considered as a complex nonlinear system.
Simple nonlinear, dynamical systems can exhibit completely unpredictable behavior, which might seem to be random. This seemingly unpredictable behavior has been called ‘chaos’. Therefore for a system to be considered nonlinear, chaos needs to be present. Essentially, the human brain is spanning on multiple scales. For example, there is established connectivity between brain networks, so the whole works better than its parts. Also, it is difficult to make long term predictions when it comes to brain function which is a hallmark for chaotic system, and the brain is all about dynamic activity suggesting varying nature of the underlying system.
What is neuronal synchronization? List two linear and two nonlinear measures of synchronization usually applied to EEG signals
Neural synchronization refers to the integration of distant brain networks to exchange information and mutually reinforce and influence one another. Therefore activity in one neural population influences another and their dynamics become mutually entrained.
Some linear measures of EEG signals are linear correlation, coherence, granger causality and multivariate modeling. Some nonlinear measures of neural synchronization are nonlinear correlation, phase-synchrony and generalized synchrony
What are three types of connectivity in brain? Discuss briefly how they are related.
There are three types of connectivity in the brain
1. anatomical connectivity – describes the physical connectivity between local circuits and distant regions
2. functional connectivity – refers to the functional integration of brain structures in terms of their statistical interdependence
3. effective connectivity – describes the causal effects of one neural system over another
These are interconnected as such as the functional and effective connectivity are constrained by structural connectivity. The structural inputs and outputs of cortical region will determine the extent of its functional connectivity. Also, to some extend functional connectivity may influence the structural connectivity by the development of new synaptic connections especially during developmental stages.
Name the three core measures to characterize a brain network. Discuss how these measures could distinguish major types of brain networks.
The three core measures to characterize a brain network are degree distribution, clustering coefficient and the characteristic path length. The degree distribution is an important quality of a network as it gives information about the distribution of local nodes, while the clustering coefficient gives information about the local structure (e.g. a measure of resilience if a node is lost), while the characteristic path length indicated the integration in brain networks and the efficiency of information transfer. Based on these features an interference can be drawn as to what type of network is being observed.
List five common network features often associated with healthy human brain.
A healthy brain will exhibit: High clustering coefficient Short-characteristic path length Hierarchical modularity Scale-free degree distribution Presence of interconnected hubs