Post on 23-Dec-2021
transcript
Electronic Neuron Model
Chapter 10
Membrane Modeling
• Nernst & Goldman equation – Resting potential
• Cable model of axon: – General cable equation
– Subthreshold response & pulse propagation
• Parallel conductance model – Behavior during activation
– Conductance variation
• Strongly tied to the concepts of electronic circuits
Physical Realization
• Realize physically the equivalent circuits
1. Analysis to verify model
– Really behave as same as the excitable tissue
– Improve understanding
– Adjust properties of the model
2. Constructing electronic circuits
– Whose behavior similar with real tissue
– Information processing similar with nature • Neuro-computing
cf: computer simulation
Classification of Neuron Model
• Based on structure of model
– Mathematical, Imaginary construction by physical laws, Physical model
• In conceptual dimensions
– Structure, Function, Evolution, Position in hierarchy
• According to physiological level
– Intraneuronal, Single neuron, Synapse, Neural interaction, Psychophysiological
• According to model parameters
– Resting, Stimulus, Recovery, Adaptation
Membrane Model
• Electronic realization of membrane excitation mechanism
– Theoretical model of Hodgkin & Huxley model
• Circuit modeling for conductance
– Between two nodes: inside & outside
– By active filters with transistors
• Parameters modification by variable resistors
• Voltage multiplied by 100: (10mV 1V)
– Other quantities in original values
Lewis Membrane Model
Block diagram of the Lewis membrane model
Circuit for potassium conductance
Circuit for sodium conductance
Response with Lewis Model
Complete Lewis membrane model
Single action pulse
A series of action pulse
Roy Membrane Model
• Simplicity than accuracy
• Neurofet
– Simplified with FET for conductance simulation • Easy implementation of amplifier with FET
Response with Roy Model
• Reasonably close the experimental results
Lewis Neuron Model
• Inclusion of excitatory & inhibitory synapse
Responses
Sodium & potassium current
Lewis
H &H
Lewis
H &H
Peak Na+ current
Steady state K+ current
Action pulse & corresponding ion currents
• Very similar with H-H model • Approximate within a order
Harmon Neuron Model
• Too complex to simulate neural networks
– Internal construction is not important
• Simplified pulse generation with multivibrator – Excitatory/inhibitory
– Drive up to 100 neurons
• Investigated 7 properties of neuron
Properties of Harmon Model
Properties of Harmon Model
Pulse obeys all-or-none law Width varies with frequency in some degree
Time from stimulus onset to output Fn of integration & refractory period
Response to constant input voltage
Propagation Model
• Inclusion of axial resistance
– Electronic realization of linear core-conductor model
6-unit chain
10-unit ring
Simulate pulse propagation in squid axon 17m/s (14~23m/s in experiments)
IC Realization
• Electronic neuron model in large quantity
– Electronic neuron as processing elements
• Stefan Prange model(1988,1990)
– Neuron with 8 synapses, with 300 transistors
• Misa Mahowald model(1991)
– CMOS and VLSI technology
– Simulated spikes in neocortical neurons accurately
– 0.1mm2 with 60uW power dissipation
– 100~200 neurons in 1cm1cm die