fMRI analyses Flashcards
Functional connectivity
Look at communication between regions –> relationship between functional activity of different brain regions
- receive input
- give output
- communicate
- before signals reach FFA, they have gone through retina, thalamus, primary visual cortex, etc.
–> System of networks
Correlations in brain activity
Simplest index functional connectivity: correaltion between activity of 2 regions across time
- seed voxel/region
Seed voxel/region
Seed voxel/region and its time-varying signal correlated with signal of all voxels of the brain => network of functionally connected voxels
Time scale of communication
Time scale of neural communication is in millisecond –> electrophysiological imaging
- time scale measured with hemodynamic imaging is slower so we also use a low-pass filter 0.1Hz
Interpretation of correlations in brain activity
- Direct influence
- mediated influence (goes through middle man)
- shared influence (direct and mediated combined –> communication to multiple locations)
–> differentiation is difficult - be aware of confounding factors
Scrubbing
Exclude time points with most motion
–> BOLD signal changes are correlated with subject motion
Correlation and directionality
Functional connectivity vs. effective connectivity
- Correlations –> psychophyciological interactions –> structural equation modelling –> dynamic causal modelling
Effective connectivity
Causal models, not real causal inference
–> model based approach
- hypothesis-driven
Functional connectivity
Observational approach -> hypothesis free
Psychophysiological interaction PPI
Examine influence of 1 region of interest on any other part of brain as function of psychologial context
- aims to identify regions whose activity depends on interaction between psychological factors and physiological factors
PPi effect
Context-specific change in relationship between brain regions
- reflected by context-dependent difference in the slope of the regression between two regional time series
Structural Equation Modelling SEm
Estimation of causal influence of multiple areas on each other, using a priori anatomical information
- VIS: primary visual cortex
- SENS: number-sensitive area
- SEL: number-selective area
Direct vs. indirect input from visual representations to number-selective representations
–> relative importance of 2 pathways modulated by format
Functional is not anatomical connectivity
Functional connectivity can change <-> anatomical connectivity is static during experiment
- if functional connectivity = direct influence of A to B, then there is anatomical connectivity
- if functional connectivity reflects mediating influence, then anatomical connectivity may not be present
Resting state fMRI RS-fMRI
Instruction to participant to rest and not think of anything in particular
- only 1 or 2 scans of 8 minutes each: very feasible even in vhallanging patient population
- Principal component analysis
Principle Component Analysis PCA
Identify components that explain most of the variance in the data
- subset of regions with high correlations in activity –> summarized by one component
Independent component analyses ICA
Data driven -> blind source separation method
Assumption: fMRI data is linear mixture of statistically independent sources with the goal to separate sources given the mixed data
Resting state networks
- resting connectivity
- default mode network
- somatomotor network
- visual network
- language network
- dorsal and ventral attention network
- frontoparietal control network
–> look at slide 14 to see the locations of these networks
Graph analysis
Paramters that summarize network properties
- high and low clustering (how many connections)
- hub: connectionpoint between 2 clusters
- integration: path length and distance
- segregation: clustering coëfficiënt
–> can see the effect of damage, for example if the hub is damaged, then the connection between is lost
Multi-voxl pattern analyses MVPA
Multi-voxel or multivariate
- voxel-wise or univeriate, in which nearby voxels are treated individually or expected to show similar signal (spatial smoothing)
- here interested in differences between voxels
Correlation MVPA
- representational similarity analysis RSA
Test whether within-condition correlation between datasets is higher than between-condition correlation –> if so: region differentiates between conditions
Decoding MVPA
Across-voxel activity patterns in dataset 1 used to train a pattern cassigier
- classifier finds a decision boundary in the multi-dimensional input space
- cross-validation: test classifier performance on independent dataset 2
Repetition suppression/adaption
- neural response decreases if stimulus repeats = adaption
- basic measure fMRI adaption: difference between repeat stimulus and different stimulus
–> asumption: areas with diminished response for repeat are sensitive to stimulus features
fMRI adaptation
slide 26 idk what he was saying tbh
