Brain Functional
Connectivity
with functional magnetic
resonance imaging"
Enrico Glerean (MSc), Brain & Mind Lab, BECS, Aalto University"
www.glerean.com | @eglerean | becs.aalto.fi/bml | [email protected]"
2 Questions"
WHY DO WE STUDY BRAIN NETWORKS?"
HOW DO WE STUDY BN WITH FMRI?"
Enrico Glerean"
www.glerean.com | @eglerean"
Why do we want to study brain
connectivity?!
• The brain is a network
~10^10 neurons and ~10^4 connections per neuron!
• As for genomics in the 20th century, many authors are
now praising the connectomics as the current revolution
in neuroscience!
• Multi-million projects like the Human Connectome
Project, the BRAIN initiative"
• If the brain is a network we want to model it as a
network!
! !
Enrico Glerean"
www.glerean.com | @eglerean"
Three important References
" Network modelling methods for FMRI "
" Smith S.M. et al.; Neuroimage 2011,!
! doi:10.1016/j.neuroimage.2010.08.063!
! Complex brain networks: graph theoretical analysis
of structural and functional systems"
! Bullmore, E, Sporns, O; Nature Rev. Neuroscience 2009
doi:10.1038/nrn2575!
" Networks of the Brain
Sporns, O; 2010, MIT Press. !
Enrico Glerean"
www.glerean.com | @eglerean"
Connectivity in
neuroscience"
CONCEPTS"
Enrico Glerean"
www.glerean.com | @eglerean"
Connectivity in neuroscience!
• Structural connectivity
(estimating actual connections)"
- Invasive (tract tracing methods, 2 photon calcium imaging)!
- Non invasive (Diffusion Tensor and Diffusion Spectral Imaging)!
• Functional connectivity
(based on temporal “co-variance”)"
- Invasive (intracranial recordings)!
- Non invasive (fMRI, M/EEG, simulated data)!
Enrico Glerean"
www.glerean.com | @eglerean"
fMRI: From univariate to multivariate
analysis
• We measure multiple time
series at once!
• We can consider them
independently (e.g. GLM) or
we can look at mutual
relationships"
Blood Oxygen Level signal"
Enrico Glerean"
www.glerean.com | @eglerean"
Network Science"
CONCEPTS"
Enrico Glerean"
www.glerean.com | @eglerean"
A (complex) network, a graph!
Newman, M. E. J., Networks: An introduction. Oxford University Press, Oxford,
March 2010.!
Enrico Glerean"
www.glerean.com | @eglerean"
Functional connectivity is multivariate!
• Functionally connected = there is a relationship
between two or more voxel time series!
• Pairwise (bivariate / multivariate) = we consider two
time series and compute their relationship to build a
network. Repeat for all pairs and then use multivariate
approaches e.g. network science."
• Multivariate = consider multiple voxels at once (PCA,
ICA, MVPA)"
Enrico Glerean"
www.glerean.com | @eglerean"
Connectivity with
fMRI"
PARADIGMS AND METHODS"
Enrico Glerean"
www.glerean.com | @eglerean"
Paradigms for functional connectivity
• Resting state FC
Looking at spontaneous BOLD activity while the subject is
in the scanner
Correlated with anatomy!
• Task related FC
The subject is performing a task with multiple conditions
(usually block design or naturalistic design, i.e. a block
design with longer blocks)!
Enrico Glerean"
www.glerean.com | @eglerean"
Building a functional network!
At each node (= voxel or region) we measure a time
series.
b1(t)!
b2(t)!
Enrico Glerean"
www.glerean.com | @eglerean"
Building a functional network!
The value of similarity between the two time series is used
as weight of the edge between the two nodes!
r12! b1(t)!
e.g. Pearsonʼs correlation:!
r12! r12 = corr(b1(t),b2(t))!
b2(t)!
Many other methods for similarity, see Smith 2011, Neuroimage!
Enrico Glerean"
www.glerean.com | @eglerean"
Building a functional network!
Repeat for all pairs of nodes and we get the full
functional network"
Enrico Glerean"
www.glerean.com | @eglerean"
Quantifying results"
SEED BASED ANALYSIS"
NETWORK ANALYSIS"
Enrico Glerean"
www.glerean.com | @eglerean"
Seed based analysis!
• Connectivity from one seed to rest of the brain!
• Usually a sphere in a selected region based on literature"
• You will end up with one correlation map (i.e. one
volume of values between -1 and 1) per subject per task,
and then you look at group or task differences (t-test,
permutation tests, etc) as done in general linear model!
Figure from: Fox, M. D. et al
(2005). The human brain is
intrinsically organized into
dynamic, anticorrelated
functional networks. PNAS!
Strongest correlations between
a seed in PCC and the rest of
the brain are shown.!
© 1970 Monty Python"
Network analysis!
• First build the network by looking at functional similarity
for each pair of nodes"
• Obtain one network per subject or per task!
• Extract meaningful information from the networks with
tools from Network science and graph theory"
Enrico Glerean"
www.glerean.com | @eglerean"
Source: Rubinov & Sporns 2010, Neuroimage.!
Network science"
EVERYTHING YOU NEED TO KNOW ABOUT
NETWORK SCIENCE CONDENSED IN FEW
MINUTES"
Enrico Glerean"
www.glerean.com | @eglerean"
Building the network!
• First step is to threshold the network to obtain a sparse
graph!
• Thresholding based on statistics or percentage of links"
b1(t)! r12! b2(t)!
threshold!
Enrico Glerean"
www.glerean.com | @eglerean"
Microscopic (node level) measures!
• Node degree/strength"
" How strong is a node?!
• Clustering"
! How close is the node !
! with the neighbours?"
• Closeness centrality"
" How distant is the node?!
• Betweenness centrality"
! How many shortest paths
through the node?"
Enrico Glerean"
www.glerean.com | @eglerean"
Microscopic (node level) measures!
• Node degree/strength"
" How strong is a node?!
• Clustering"
! How close is the node !
! with the neighbours?"
• Closeness centrality"
" How distant is the node?!
• Betweenness centrality"
! How many shortest paths
through the node?"
Enrico Glerean"
www.glerean.com | @eglerean"
Mesoscopic (module level) measures!
• Communities/clusters"
" Finding subset of nodes
that are forming a module!
• Comparisons of
modularity"
" How different brains have
different (sub)networks
Fortunato, S. (2010). Community detection in graphs.
Physics Reports, 486(3-5), 75–174!
Alexander-Bloch, et al. (2012). The discovery of
population differences in network community
structure NeuroImage, 59(4), 3889–900. !
Enrico Glerean"
www.glerean.com | @eglerean"
Global measures!
• Clustering"
! How densely connected
are the nodes?!
• Path length"
" Average distance between
nodes"
• “Small world-ness”
Many short distance links
(high clustering) and few
important long distance ones
(short path)!
Enrico Glerean"
www.glerean.com | @eglerean"
Quantifying network results: statistics!
• Micro/meso/macro scopic measures can be compared
across groups or conditions with standard tests!
• Correlation between measures "
• Non independency of values"
• Permutation based tests "
Enrico Glerean"
www.glerean.com | @eglerean"
Examples from
decision-making
literature"
Enrico Glerean"
www.glerean.com | @eglerean"
Examples from decision-making
studies with fMRI and FC!
• Minati et al (2012) Decision-
making under risk: A graph-
based network analysis
using functional MRI.
Neuroimage doi:10.1016/j.neuroimage.
2012.02.048!
- A network of inter-regional connections
that are modulated by the expected value
of a gambling task; main hubs in the PFC!
• Li et al (2013) Resting-State Functional Connectivity
Predicts Impulsivity in Economic Decision-Making
Journal of Neuroscience doi:10.1523/JNEUROSCI.1342-12.2013!
Enrico Glerean"
www.glerean.com | @eglerean"
Artefacts and
confounds"
EFFECTS OF HEAD MOTION AND NOISE"
Enrico Glerean"
www.glerean.com | @eglerean"
Sources of noise for FC!
• Head motion"
- Effects of head motion on FC carefully assessed by a recent paper
Power, J. D et al. (2014) Methods to detect, characterize, and remove motion
artifact in resting state fMRI. NeuroImage. doi:10.1016/j.neuroimage. 2013.08.048!
• Scanner and physiological (heart, breathing) noise"
- the same noise in two time series will cause them to be similar!
- de-noise the data or at least band-pass filter it (0.01 – 0.08 Hz)!
Source: Glerean et al 2012, Brain Connectivity doi:10.1089/brain.2011.0068!
Tools and
resources"
Enrico Glerean"
www.glerean.com | @eglerean"
Interesting toolboxes!
• Generalized PPI toolbox"
" http://www.nitrc.org/projects/gppi!
• CONN toolbox
http://www.nitrc.org/projects/conn/!
• Functional Connectivity toolbox
https://sites.google.com/site/functionalconnectivitytoolbox/!
• Brain connectivity toolbox
https://sites.google.com/site/bctnet/ "
• All Matlab® based but check NiTime for Python stuff"
Enrico Glerean"
www.glerean.com | @eglerean"
Interesting resources
• 1000 resting state subjects
http://fcon_1000.projects.nitrc.org/!
• Human Connectome Project
http://www.humanconnectomeproject.org/!
• ADHD 200
http://fcon_1000.projects.nitrc.org/indi/adhd200/!
• Alzheimerʼs Disease Neuroimaging Initiative
http://adni.loni.usc.edu/ "
Enrico Glerean"
www.glerean.com | @eglerean"
Take home
messages"
Enrico Glerean"
www.glerean.com | @eglerean"
Three take home messages!
1) Brain networks: Modelling the brain as a network
gets us closer to the actual functioning mechanisms
of the brain and network science helps us identifying
salient features from the network"
2) Functional brain networks: fMRI – thanks to its great
spatial resolution – is a useful tool not only to look at the
intrinsic connections but also to explore how
connectivity changes with task"
3) Beware: As with every neuroimaging method, you have
to be careful to make sure that you are measuring
actual brain activity and not artefacts"
Enrico Glerean"
www.glerean.com | @eglerean"
