Functional integration : Analysis of distributed effects, it’s basically the process by which different neural systems within the brain interact functionally to integrate information across different cognitive or sensory modalities. More simply, how different brain areas work together to process information.
Dynamic phenomenon at variable time scale : You can basically have two types of studies, either you see the temporal correlations between distant neurophysiological events (fMRI time series for example) or the influence of a neural system onto another.
What are the 2 methods for observing a dynamic phenomenon at variable time scale ?
Study of time series = data driven analysis, voxel time course thanks to fMRI, no indication on how correlations were generated
Modulation of brain connectivity = is based on a model of interactions with graph (connections) and mechanisms of interaction (synapse population and others)
What are the steps for connectivity estimation ?
Acquisition
preprocessing
either functional or effective connectivity
for function : identify functional network without having an "a priori" functional model
for effective : identify the modt likely model of interaction leading or coming from one or more functional models
Functional connectivity : connection between brain areas based on their activites over time
Unlike structural connectivity, functional connectivity is highly time-dependent. Statistical patterns between neural elements fluctuate on different time scales, some as short as tens or hundreds of milliseconds
Effective connectivity : How one brain area directly affects the activity of another.
Causality = cause-effect relationship
Correlation = link without providing the cause
Bayesian model : is a statistical model that combines old data and new evidence in order to make predictions
intégration = c’est ce qui généralise la corrélation à plusieurs régions
ICA analysis : component analysis that it is a statistical way to find independent signals mixed in the data
PCA analysis : component analysis that simplifies the data and get the most important stat feature
Techniques used in preprocessing : things like normalization, realignment, smoothing and artifact removal.
Steps for resting state fMRI ?
Bold signal
Select a ROI or seed region
Need to see the correlation between seed voxel and the resting-state time-series of the region
Draw the correlation
functional correlation map
functional connectivity is sensitive to the task performed during data acquisition.
Covariance matrix is a matrix containing covariance scores, which measure how much two variables change together for each pair of variables in the dataset.
Psychophysiological interaction PPI is a method of imaging that examines relationship between brain regions changes in response to a task or external stimuli.
Pros of PPI ?
given a region you can test context-dependent connectivity in entire brain
simple to perform
based on regression
assumes dependent and indepent variables
Cons of PPI ?
very simple model : only allow modeling contribution from 1 area
ignores the time-series properties of data
Methods to use in EEG and EMG ?
Synchrony
Volumic conduction
Coherence
What is volumic conduction ?
Phenomenon where electrical signals spread through brain tissues and fluids, affecting measurements like EEG.
Coherence between 2 signals mean that you'll measure the synchronization between 2 signals based on their phase consitency
Coherence ranges from 0 to 1
What is coupling between frequencies ?
Refers to the relationship/influence between different frequency bands within the brain electrical/magnetic activity.
What are the 2 steps of coupling frequencies ?
first : identify functional network and calculate indices coherence
second : look for correlations between coherency indices and behavioral variables
what are the 3 methods for coupling ?
PPC
PAC
AAC
Structural Equation Modeling (SEM) : A statistical technique that tests and estimatescausal relationships using an approach that combines factor analysis and multiple regression
Steps of SEM ?
Anatomical model
extracting signal
calculate covariancematrix
estimate path coefficients
comparing path coefficients
DCM (dynamic causal modeling) is a framework for modeling directional interactions between brain regions and how they are influenced by external inputs.
DCM is based on a hypothesis about functionning that is clearly defined
Granger causality is a statistical hypothesis test to determine whether one time series can predict another : defined by both time + frequency domains
What is granger causality based on ?
multivariate autoregressive (AR) model
multivariate autoregressive (AR) model : method to predict a variable based on past values of itself and others
Techniques for anatomical connectivity ?
Dissection + microscopie
Traçage
Tractographie IRM par tenseur de diffusion
EEG/MEG techniques involve oscillations, synchrony and coherence
Effective connectiviy involves granger connectiivty, SEM and DCM
Studies of the brain connectome can be done at the micro, meso and macro scales
Microscale is related to individual neurons and their synaptic connections
Mesoscale encompasesses connection within and between microcolumns or another local cell assemblies
Macroscale refers to anatomically segregated brain regions and their projections