Neuroimaging simulations structural vs functional
Transcript of Neuroimaging simulations structural vs functional
Neuroimaging simulations structural vs functional
Workshop on topological data analysis
Daniela Egas Santander Fields institute, June 2020 1
Henry Markram, Taylor Newton, Max Nolte, and Michael Reimann (Blue Brain Project, EPFL)
Dejan Govc, Jānis Lazovskis, Ran Levi, Daniel Lütgehetmann, Jason Smith (Aberdeen)
Stefania Ebli, Kathryn Hess, Etienne Lasalle, Nicolas Ninin, Gard Spreeman (Laboratory for Topology and Neuroscience, EPFL)
Nicolas Antille and Jonas Karlsson (visualization)
Collaborators
2
Biological neural networksNeural systems
Networks
Structural networks Functional networks
Physical/material Statistical: (generalized) correlations between time series
➢ Spike trains
➢ Neuroimaging (fMRI, VSD, calcium imaging)
3Today explore connections
Estimated connectivityFunctional connectivity increasingly popular Question: High correlations neural communication?
“ Structure shapes function”
Spike trains
Lab work
Neuroimaging
Few neurons observed
Structural connectivity not measurable
Problem!
Too little ground truth
4
how?
Digital reconstruction of the somatosensory
cortex of a rat
~ 31 000 neurons
~ 8 million synaptic connections
Numerous experiments in the lab digitally
reproduced, without parameter tuning
The Blue Brain Microcircuit
5
Voltage Sensitive Dye (VSD)
microcircuit voxelization point spread
Taylor Newton (BBP)
6
The Blue Brain as a digital laboratory
Structural network is known
~30K neurons observed
BBP
Spike trains Few neurons observed
Structural connectivity not measurable
Lab work
Neuroimaging
7
From microcircuits to topology
82D/3D -Filtered cubical complex
Functional VSD
Directed Flag Complex
Structural microcircuit
Structure shapes function
high VSD correlation
Neurons belonging to high dimensional simplices have:
Reimann, Nolte, et al., Cliques of Neurons Bound into Cavities Provide a Missing Link between Structure and Function
high spiking correlation
9
Studying activity
Time series
Transmission-response
graphs
Filtrations of a cubical complex
10
Topological signatures of activity
11Reimann, Nolte, et al., Cliques of Neurons Bound into Cavities Provide a Missing Link between Structure and Function
Homological signature of spiking activity
Structural
Persistence signature of imaging activity
Functional
Functional and structural connectivityFunctional connectivity neural communication
“Structure shapes function”Spike trains
VSD data
Structural microcircuit Functional microcircuit
topological invariants?
(complexity)
qualitative
12
Structure of the brain
β1
β3
https://github.com/luetge/flagser 13
Functional connectomes and smoothness
Functional connectome
Thresholding
Persistence
How to see more than just close neighbours?
How to detect surprisingly high correlations?
14
Semi-naive thresholding
15
Structural Functional
SummaryStructure shapes function:
1. Correlations and structural graph
2. Structural and functional swooshes
3. Functional and structural microcircuits (in progress)
16
Robust methods for thresholding or correcting distance dependence
Finite cubical complexes with random filtration
Statistical methods for the construction of random directed graphs with fixed simplex counts
Perspectives