ECE 795:

Quantitative
Electrophysiology
Notes for Lecture #3
Friday, October 3, 2008
5. LINEAR CABLE EQUATIONS

We will look at:

¾ Core-conductor model
¾ Cable equations
¾ Linear (subthreshold) response of a
cylindrical fiber

2
Core-conductor model:
In the core-conductor model we approximate an
axon or a segment of a dendrite as a uniform
cylinder.

(from Johnston
and Wu)

Each small (cylindrical) segment of membrane is

electrically linked (axially) to the next segment by
the intra- and extra-cellular electrolytes. 3
Resistance and capacitance in a cylindrical
fiber:
If the resistivity of the intracellular electrolyte is
Ri (Ω·cm), then for a cylindrical fiber of radius a the
axial (longitudinal) resistance per unit length is:

(Note the convention that (i) resistivity or specific

resistance/capacitance is designated by an
uppercase letter and (ii) resistance or capacitance
per unit length is designated by a lowercase letter.)
4
Resistance and capacitance in a cylindrical
fiber (cont.):
If Rm (Ω·cm2) and Cm (μF/cm2) are the specific
resistance and the specific capacitance,
respectively, of the membrane, then the
membrane resistance times length is:

and the membrane capacitance per unit length is:

5
Core-conductor model (cont.):
If a single fiber described by the core-conductor
model lies in a restricted extracellular space, then
longitudinal current flow can occur in the
extracellular electrolyte and longitudinal variations
in the extracellular potential can result.

6
Core-conductor model (cont.):
Under linear (i.e., subthreshold) conditions, each
membrane patch of length Δx can be described by
a lumped RC circuit.

7
Core-conductor model (cont.):
Under nonlinear (i.e., suprathreshold/
transthreshold) conditions, each membrane patch
of length Δx must be described by the HH model.

8
Core-conductor model assumptions:
1. The transmembrane and longitudinal currents,
as well as the intra- and extra-cellular
potentials, are functions only of the axial
(longitudinal) coordinate x. That is, we have a
one-dimensional cable model.
2. For a fiber with a restricted extracellular space,
the extracellular current can only flow in the
axial (longitudinal) direction.
In the case of a larger extracellular space, the
resistance of the extracellular electrolyte is
assumed to be negligible, i.e., re ≈ 0.
9
Nerve fiber bundle showing restricted extracellular
spaces:

10
Core-conductor model assumptions (cont.):
3. The radius of a fiber is typically many times
smaller than its length, such that the
intracellular current can be assumed only to
flow in the axial (longitudinal) direction. The
resistance per unit length of the intracellular
electrolyte is found via Eqn. (2.550).
4. For nerve and muscle under passive
conditions, the membrane is represented by
passive components (shown in Fig. 6.2) with
values for rm and cm found via Eqns. (2.560)
0
and (2.57 ).
Under active conditions, the HH model is
utilized, as illustrated in Fig. 6.3. 11
Cable equations:
Ohm’s and Kirchhoff’s laws can be applied
to the core-conductor circuit (shown in Fig.
6.1 of Plonsey and Barr) to obtain the cable
equations for a uniform cylindrical fiber of
arbitrary length.
It is then desirable to evaluate the cable
equations in the limit as Δx → 0, such that
the equations describe the behaviour of the
fiber as a continuous function of axial
(longitudinal) position, rather than a set of
discrete membrane patches. 12
Cable equations (cont.):
The resulting relationship between the extracellular
potential gradient in the axial direction as a
function of the axial current is:

and likewise the intracellular potential gradient is:

13
Cable equations (cont.):
If current leaves the intracellular space by crossing
the membrane, then the intracellular current will
show an axial decrease, while the transmembrane
current will be positive. This conservation of
current is described by:

where im is the transmembrane current per unit

length.
(Note that im is a linear function of the membrane potential
under passive (subthreshold) conditions but is a nonlinear
function under active (suprathreshold) conditions.) 14
Cable equations (cont.):
In contrast, the extracellular current will increase
axially due to any transmembrane current.
In stating this relationship we will also allow for the
possibility that a current may be injected into the
extracellular space from polarizing electrodes,
giving:

where ip is the current per unit length injected from

the polarizing electrodes.
15
Cable equations (cont.):
Example current pathways for extracellular current
injection:

16
Dependence of membrane current on Vm:
Suppose I is defined as:

From Eqns. (6.3) and (6.4):

That is, any change in the total axial (longitudinal)

current must come from the injected current ip. 17
Dependence of im on Vm (cont.):
We now consider the relationship between the
transmembrane potential and the extra- and intra-
cellular currents and potentials.
Since, by definition Vm = Φi − Φe, we have:

18
Dependence of im on Vm (cont.):
If Eqns. (6.8) is differentiated with respect to x,
then:

Substituting Eqns. (6.3) and (6.6) gives:

19
Dependence of im on Vm (cont.):
In comparison, if Eqn. (6.2) is differentiated with
respect to x and Eqn. (6.3) is substituted for
∂Ii/∂x, then:

¾ Note that Eqn. (6.12) shows the dependence of the

transmembrane current im on the transmembrane
potential, the injected current and the extra- and intra-
cellular resistances.
¾ In contrast, Eqn. (6.13) describes the dependence of
the transmembrane current on the intracellular
potential and resistance only. 20
Potentials φi and φe from vm:
From the cable equations, it is possible to work out
how a change in the transmembrane potential is
split between the changes in the intra- and extra-
cellular potentials:

21
Potentials φi and φe from vm (cont.):
If no source exists within the region (x,∞), i.e., ip
= 0, then I = 0 in that region and the integrals in
Eqns. (6.20) and (6.22) drop out.
In such a source free region, the extra- and intra-
cellular potentials are related to the
transmembrane potential via voltage divider type
expressions:

22
Linear (subthreshold) response of a
cylindrical fiber:
Under subthreshold conditions, the
transmembrane current per unit length
im (mA/cm) in a cylindrical fiber is:

where rm is the membrane resistance times

unit length (Ω·cm) and cm is the membrane
capacitance per unit length (μF/cm).
23
Linear (subthreshold) response of a
cylindrical fiber (cont.):
Substituting Eqn. (7.12) into cable equation
(6.11) gives:

where:

24
Linear (subthreshold) response of a
cylindrical fiber (cont.):
For steady-state conditions (∂vm/∂t = 0),
Eqn. (7.14) simplifies to:

In the case of ip = 0, Eqn. (7.16) becomes:

25
Linear (subthreshold) response of a
cylindrical fiber (cont.):
The solution of Eqn. (7.17) is:

where A and B are constants, the values of which

are determined by boundary conditions.
Note that in cases where ip ≠ 0, rather than solving
Eqn. (7.16) for the particular solution, is possible to
apply Eqn. (7.17) to regions of a fiber where ip = 0
and use the region where ip ≠ 0 to impose the
boundary conditions and solve for A and B.
26
Space and time constants:
In Eqn. (7.14) we introduced:
¾ the space constant λ, and
¾ the time constant τ.

Space constant:-
Under steady-state conditions, Eqn. (7.18)
describes the space constant λ as the
distance over which the transmembrane
voltage and current decay by the factor 1/e.
27
Space and time constants (cont.):
For a cylindrical fiber with uniform membrane
properties and with re ≈ 0:

Substituting for ri and rm in Eqn. (7.19) using

0 0
Eqns. (2.55 ) and (2.56 ) gives:

28
Space and time constants (cont.):
Time constant:-
For an isopotential patch of membrane (e.g., a
small spherical cell), we saw in Eqn. (7.3) that the
time constant τ corresponds to the time over which
the transmembrane potential grows towards its
steady-state value by the factor 1−1/e.
For a cylindrical fiber, the transmembrane potential
grows to a particular fraction its steady-state value
by the time τ, where the fraction depends on the
distance from the site of stimulation.

29
Steady-state solution for stimulus current at
the origin:
Injection of a small current into the extracellular
space at the origin (center) of an infinitely-long
cylindrical fiber can be approximated by a spatial
delta function source:

where I0 is the total applied current (mA) and δ(x)

is the unit delta function.
Note that ip is zero everywhere except at the
origin, where it is infinite, and integrating ip around
the origin gives the total current I0.
30
Steady-state solution for stimulus current at
the origin (cont.):
Given the current injection described by
Eqn. (7.22), the steady-state membrane
equation (7.16) becomes:

Except at the origin, the fiber is described by

the homogeneous equation (7.17), which
yields the homogeneous solution given in
Eqn. (7.18).
31
Steady-state solution for stimulus current at
the origin (cont.):
However, vm(x) must be continuous at the
origin, so Eqn. (7.18) must apply
everywhere.
Now we need to apply the boundary
conditions of:
¾ a source at the origin x = 0, and
¾ the transmembrane potential at |x | = ∞,
to determine the constants A and B.
32
Steady-state solution for stimulus current at
the origin (cont.):
¾ The relative transmembrane potential as x →
∞ must be zero, so the constant B must be
zero for the region x ≥ 0.
¾ Likewise, the relative transmembrane potential
as x → −∞ must be zero, so the constant A
must be zero for the region x ≤ 0.
¾ Because of the continuity of vm(x) at the origin,
the value of the constant A for the region x ≥ 0
must be equal to the constant B for the region
x ≤ 0. We will call this constant C.
33
Steady-state solution for stimulus current at
the origin (cont.):
These boundary conditions can be summarized
as:

Given these boundary conditions, Eqn. (7.18) can

be rewritten as:

34
Steady-state solution for stimulus current at
the origin (cont.):
Integrating Eqn. (7.30) around the origin and
imposing these boundary conditions to solve for
the constant C, we obtain:

Substituting Eqn. (7.33) into Eqn. (7.30) gives:

35
Steady-state solution for stimulus current at
the origin (cont.):
Inspection of Eqn. (7.34) leads to the following
conclusions:
1. The stimulus clearly affects the
transmembrane potential, since vm is nonzero
for all finite values of x.
2. The effects of the stimulus varies markedly
with x. The largest change in vm occurs where
x = 0. vm decreases exponentially with
distance, falling by a factor of 1/e every length
λ from the stimulus site.
36
Steady-state solution for stimulus current at
the origin (cont.):
3. For a given stimulus I0, the change in vm
increases with increasing re, increasing rm
and/or decreasing ri.
4. From the sign of Eqn. (7.34), a positive current
injected into the extracellular space leads to a
negative vm, i.e., the membrane is
hyperpolarized under an extracellular anode.
5. The space constant λ gives a nominal
measure of how far the disturbance in vm
extends from the site of stimulation.
37
Step current at origin – general time-varying
solution:
Consider the homogeneous version of Eqn.
(7.14):

As before, we will apply the boundary condition of

an injected current at the origin.
If we assume an unbounded extracellular space
and an intracellular injected current, we can
assume that re ≈ 0. 38
Step current at origin – general time-varying
solution (cont.):
To simplify our notation, we introduce the
normalized space and time variables:

such that Eqn. (7. 35) becomes:

39
Step current at origin – general time-varying
solution (cont.):
The solution to Eqn. (7.37) for an infinite
cable with an intracellularly-injected current
I0 at the origin can be obtained by using:
¾ the Laplace transform to turn the
differential equation into an algebraic
equation, and
¾ the boundary conditions of the injected
current at X = 0 and vm(X) = 0 at
X = ±∞.
40
Step current at origin – general time-varying
solution (cont.):
In the normalized space and time variables
the solution is:

where:

41
Step current at origin – general time-varying
solution (cont.):
Converting back to the original coordinates
gives:

42
Step current at origin – general time-varying
solution (cont.):
For a given value of time, the spatial
behaviour is exponential-like. For t > τ,
vm(x) tends towards a true exponential, as
was obtained for the steady-state response
described by Eqn. (7.34).
This continuous decrement of vm(x) with
increasing |x| is due to the leakage of
current through the membrane, while λ
describes the rate of this effect.
43
Step current at origin – general time-varying
solution (cont.):
For a given position x along a fiber, the
membrane potential reaches its steady-state
in an exponential-like manner over time.
Only at x = λ is it truly
exponential, i.e., the
fraction of the
steady-state potential
that is achieved
at t = τ is 1−1/e.
44
Step current at origin – general time-varying
solution (cont.):

45
Step current – time-varying soln. (cont.):

46
6. PROPAGATION OF ELECTRICAL
POTENTIAL WAVEFORMS
We will look at:
¾ Axonal delays & propagation velocity in linear
cable
¾ Local circuit currents during propagation
¾ Mathematics of propagating action potentials
¾ Numerical solutions for propagating action
potentials
¾ Propagation velocity constraint for uniform
fiber
¾ Propagation in myelinated nerve fibers
47
Impulse propagation:
In the previous lecture we considered
excitation and action potential generation in
an isopotential patch of membrane.
However, in practice we are often concerned
about the propagation of transmembrane
potential impulses (i.e., waveforms),
particularly action potentials, along the
length of axons, dendrites or muscle fibers.

48
Impulse propagation (cont.):

(from Johnston
and Wu)

49
Axonal delays & propagation velocity in
linear cable:
Figure 7.4 on slide # 45 shows a delay in
depolarizing sites far away from the site of current
injection. Is this delay somehow proportional to
the distance, i.e., can one define a propagation
velocity?
There is no wave solution for a linear cable:
1. waveforms must dissipate due to the low-pass
character of the linear membrane, and
2. there must be an infinitely-quick response at
distant sites if the cable contains no inductive
elements. 50
Axonal delays & propagation velocity (cont.):
However, it is possible to define a propagation
delay/velocity based on the centroid (first moment)
of potentials in a linear cable. For a current or
voltage waveform h(x,t), the centroid at location x
is:

51
Axonal delays & propagation velocity (cont.):
The transfer delay D is the difference between the
centroid of the induced voltage measured at
location y and the centroid of the current that was
injected at location x:

The local or input delay is then Dxx.

The propagation delay P is the difference in the
voltage centroids at x and y:

52
Axonal delays & propagation velocity (cont.):
These delays can be obtained for a linear cable by
multiplying the cable equations by t and integrating
over t. The resulting equation is an ODE similar to
the steady-state cable equations.
Some important properties are:
1. Dxy > 0
2. Dxy is independent of the form of Iinj, i.e., it is a
properties of the cable not the input, and
3. Dxy = Dyx.

53
Axonal delays & propagation velocity (cont.):
For an isopotential neuron, Dxx = τm.
For an infinite cable, Dxx = τm/2,

and

such that the nominal propagation velocity is:

54
Axonal delays & propagation velocity (cont.):

(from Koch)

55
Local circuit currents:
Propagation of action potentials can be understood
qualitatively by considering the patterns of local
currents that are
produced by an action
potential (site A in the
figure below).

56
Local circuit
currents (cont.):

(from Koch)

57
Mathematics of propagating action potentials:
The cable equations describe the behaviour of
the extra- and intra-cellular (and consequently
transmembrane) currents and potentials as a
function of space, specifically the axial
(longitudinal) coordinate x.

In order to describe the propagation of

transmembrane potentials, i.e., movement in
space over time, we must couple the cable
equations with a description of how a patch of
membrane behaves as a function of time, i.e., the
linear (passive) RC circuit equation or the
nonlinear (active) HH model equations.
58
Mathematics of propagating action
potentials (cont.):
Consider the membrane current equation
generalized to axial position x on a uniform cable:

Because of the nature of the capacitive current,

Eqn. (6.25) can be reformulated to give:

59
Mathematics of propagating action
potentials (cont.):
The transmembrane current per unit area, Im, is
related to im (the current per unit length) via the
cylindrical geometry, such that:

and consequently from Eqn. (6.12):

60
Mathematics of propagating action
potentials (cont.):
Eqn. (6.27) gives the partial derivative of the
membrane potential with respect to time t,
whereas Eqn. (6.31) gives the (2nd) partial
derivative with respect to space x.
A single PDE describing the membrane potential
behaviour in time and space can be obtained by
substituting Eqn. (6.31) into Eqn. (6.27):

61
Mathematics of propagating action
potentials (cont.):
To model the propagation of action potentials, the
ionic current term from Eqn. (6.27) can be found
from the HH model current equations:

62
Numerical solutions for propagating action
potentials:
¾ The equation given on slide 5 is a nonlinear
partial differential equation (PDE) that is first-
order in time and second-order in space.
¾ Numerical solution is required because of the
nonlinear nature of the ionic current given by
Eqn. (6.28).
¾ The fiber must be discretized into sequential
points (or nodes of Ranvier) spaced at intervals
of Δx (say 25 μm).
¾ The solution is then computed for each time
interval Δt (say 10 μs).
63
Numerical solutions for propagating action
potentials:
One method for numerical solution is to compute
these four steps at each time interval:
1. Solve for the change in transmembrane
potential ΔVm by approximating the partial
derivates for discrete intervals.
2. Update Vm.
3. Compute the updated gating particle time
constants αn, βn, etc.
4. Compute the new ionic currents based on the
updated gating time constants and Vm.
This has updated Iion and Vm by one time step. 64
Numerical solutions for propagating action
potentials (cont.):
Note that it is often desirable to have a variable
time step Δt, to optimize computation speed.
This can be achieved with:
¾ Matlab, which has a set of numerical PDE
solvers, or
¾ software packages for simulating axons and
dendrites, including:
– NEURON: For computer simulations of
neurons and neural networks
– The GEneral NEural SImulation System
(GENESIS) 65
Example propagating action potential:

66
Example propagating action potential
(cont.):

67
Example propagating action potential
(cont.):

68
Propagation velocity constraint for uniform
(unmyelinated) fiber:
For uniform propagation, the space-time
behaviour of Vm(x,t) must satisfy the wave
equation:

where θ is the propagation velocity and the

waveform for Vm on the RHS is the temporal
waveform, i.e., Vm(t).
69
Propagation velocity constraint for uniform
(unmyelinated) fiber:
Alternatively:

where the waveform for Vm on the RHS is

the spatial waveform, i.e., Vm(x).
(Note that propagation of the action potential
waveform without dissipation requires an
active membrane, i.e., voltage-gated ion
channels.) 70
Propagation velocity constraint for uniform
(unmyelinated) fiber (cont.):
Differentiating Eqn. (6.64) once with respect
to x, utilizing the chain rule, gives:

and again results in:

71
Propagation velocity constraint for uniform
(unmyelinated) fiber (cont.):
Substituting Eqn. (6.66) into (6.31), with
ip = 0, re = 0 & ri = Ri/πa2, gives:

With an appropriate value for θ the solution to this

ODE exhibits an action potential; the solution
diverges with an incorrect θ. 72
Propagation velocity constraint for uniform
(unmyelinated) fiber (cont.):
An important property of the propagation
velocity can be obtained with inspection of
Eqn. (6.68) without solving it explicitly.
Note that all the terms on the right-hand side
(6.68) are independent of the fiber radius a,
as is d2Vm/dt2, and thus the coefficient must
be a constant independent of a, that is:

73
Propagation velocity constraint for uniform
(unmyelinated) fiber (cont.):
Consequently:

Since K is unknown, it must be determined

experimentally.
An empirically-obtained relationship for squid giant
axon at 18.3°C is:

where d is the fiber diameter in μm. 74

Propagation in myelinated nerve fibers:
In vertebrates, Schwann cells produce myelin
which wraps around an axon to produce an
insulating sheath. The regularly-space breaks in
the myelin are called nodes of Ranvier, and the
axon segments between nodes are referred to as
internodes.

75
Propagation in myelinated nerve fibers
(cont.):
The myelin is wrapped in layers around the axon,
often on the order of 10s or even 100s of layers.

76
Propagation in myelinated nerve fibers
(cont.):

(from Koch)

77
Propagation in myelinated nerve fibers
(cont.):

(from Koch) 78
Propagation in myelinated nerve fibers
(cont.):
The specific leakage resistances and
specific capacitances of the myelin sheath
and cell membrane shown below are
consistent with the myelin sheath being
equivalent to around 100 layers of cell
membrane.

79
Propagation in myelinated nerve fibers
(cont.):
Considering the Frankenhaeuser–Huxley model
under subthreshold (i.e., linear/passive) conditions,
the nodes of Ranvier have a specific membrane
resistance and specific membrane capacitance of:

respectively.
80
Propagation in myelinated nerve fibers
(cont.):
Note:-
¾ Nodes of Ranvier are around 1 μm in length.
¾ Internodal distances are on the order of 1 to 2
mm. (A rough empirical rule is that the
internodal length equals 100×d, where d is the
fiber diameter.)
Although internodes are much longer than nodes,
the much smaller specific capacitance of the
former means that an internode and a node have
approximately the same capacitance.
81
Propagation in myelinated nerve fibers
(cont.):
The purpose of the myelin is clearly to:
1. reduce the capacitance of long stretches of
membrane, the internodes, such that they do
not need to be charged up for action potential
propagation, and
2. increase the membrane leakage resistance so
that there is less leakage across the
membrane of the intracellular longitudinal
current.
Consequently, the “local circuit currents” extend
over much longer lengths of the fiber. 82
Propagation in myelinated nerve fibers
(cont.):

83
Propagation in myelinated nerve fibers
(cont.):
Because the local circuit currents extend
from node to node, action potentials
effectively jump or skip from node to node,
which is referred to as saltatory propagation.
Saltatory propagation produces:
1. faster propagation of action potentials,
and
2. a “failsafe” mechanism – if one node is
blocked, the action potential will skip
over it to the next. 84
Propagation in myelinated nerve fibers
(cont.):
In contrast to (6.72), for frog myelinated fibers:

where d is the fiber diameter in μm.

(from Koch) 85