Transcript Courtesy by Schoffelen and Fries FC Donders Center

NeuroInformatics :
Bridging the gap between neuron and neuroimaging.
Stan Gielen
Dept. of Biophysics
University of Nijmegen
26-11-04
Neuro_Informatics Workshop
Relevance for Neuro-Informatics
• Scientific topic: interpretation of neuronal data:
spike signals <=> neuro-imaging signals.
• Neuron-models
• good (complex) data
26-11-04
Neuro_Informatics Workshop
The neural code
Firing rate
Recruitment
Neuronal
assembly
26-11-04
Synchronous firing
Neuro_Informatics Workshop
Synchronization of firing related to attention
26-11-04
Neuro_Informatics Workshop
Riehle et al. Science, 1999
Hazard rate modulates motor cortical
oscillatory neuronal activity
Courtesy by
Schoffelen and
Fries
26-11-04
Neuro_Informatics Workshop
F.C. Donders
Center
Coherence between motor cortex
and muscle EMG
Courtesy by
Schoffelen and
Fries
26-11-04
Neuro_Informatics Workshop
F.C. Donders
Center
Scientific questions
• Beta (15-30 Hz) and gamma (30-70 Hz) oscillations in
EEG and MEG have poor signal-to-noise ratio:
epiphenomenon, artefact or functional significance ?
• To what extent are EEG/MEG oscillations a reflection of
common, synchronized action potential firing ?
• If neuronal synchronization has any functional
implications: what are the mechanisms to initiate it or to
stop it ?
26-11-04
Neuro_Informatics Workshop
Why exploring neuron models ?
26-11-04
Neuro_Informatics Workshop
Neuron models
Leaky Integrate-and-Fire neuron
dVi

 (Vi  70 mV )  R
dt
w I
Hodgkin-Huxley
conductance based neuron
0 mV
ij j
j
V
V mV
-Cm dV/dt = gmax, Nam3h(V-Vna) +
time
26-11-04
gmax, K n4 (V-VK ) + g leak(V-Vna)
Neuro_Informatics Workshop
Simple model
Common input
n1(t)
n2(t)
X
Y
Correlation ?
26-11-04
Neuro_Informatics Workshop
Cross-correlation Kxy (t ) between
output spikes of two conductance
based
Amount of neurons with common input
common input
i  30Hz
i  90Hz
Stroeve & Gielen (2000)
26-11-04
Neuro_Informatics Workshop
Cross-correlation Kxy (t ) between
output spikes of two conductance
based
Amount of neurons with common input
common input
i  30Hz
i  90Hz
Correlated firing is a poor measure for common input
!
Stroeve & Gielen, 2000)
26-11-04
Neuro_Informatics Workshop
Neuronal properties are not constant !
EPSP
synaptic background
activity
IPSP
Membrane conductance
No synaptic
background activity
26-11-04
background excitatory rate
Kuhn, Aertsen, and
Rotter, The Journal of
Neuroscience, 2004
24:2345–2356
Neuro_Informatics Workshop
Neuronal properties
• Depend on synaptic input
– amplitude of EPSP and IPSP
– time constant of neuron
– leaky integrator versus coincidence detector
26-11-04
Neuro_Informatics Workshop
Neuronal signals
26-11-04
Neuro_Informatics Workshop
Neuronal signals
Synaptic
contact
Local Field
Potential
Action Potential
26-11-04
Neuro_Informatics Workshop
Single-unit and
multi-unit activity
Crosscorrelation
LIF
Hodgkin-Huxley
Crosscorrelation is a poor measure for common input;
coherence is a better measure
26-11-04
Neuro_Informatics Workshop
Coherence function for 60% common input
LIF
26-11-04
Hodgkin-Huxley
Neuro_Informatics Workshop
Coherence Local-Field-Spike
at 60% common input
LIF
26-11-04
Hodgkin-Huxley
Neuro_Informatics Workshop
Spike Field Coherence
Single unit
coherence
0.3
0.1
0
26-11-04
Multi unit
0.2
20 40 60 80 100120
Neuro_Informatics Workshop
Relevance for future
NeuroInformatics research
• Data-base of neuronal models in NEURON and
GENESIS
• Data-base of complex physiological data:
– combined local field potentials and single/multi unit
recordings
– combined neuro-imaging signals and local field/action
potential recordings
26-11-04
Neuro_Informatics Workshop