Bioelectric Source Model and Brain Imaging Dezhong Yao School of Life Sci & Tech,UESTC.

Bioelectric Source Model and Brain

Imaging

Dezhong Yao

School of Life Sci & Tech,UESTC

Special Thanks to

Prof Chen

for giving the chance of the talk!

1. Bioelectromagnetic Source models

2. 2D Imaging of brain activities

3. 3D Imaging of brain activities

4. EEG reference problem

CONTENT

(2) the Extracellular Current contributes directly to the scalp (EEG)

Bioelectric Source

(3) the Extracellular Current (EEG) is due to the intracellular current (source)

(1)For a live neuron, there are two currents

Conclusion: the source of EEG is the intracellular current

(1) MEG is generated by intracellular current

Biomagnetic Source

(2) The source of both MEG and EEG is the intracellular current

(1) The bioelectromagnetic source is the intracellular current

1.Bioelectromagnetic Source models

What is the bridge from current to charge or dipole model?

(2) The conventional source model is such as charge 、 dipole 、 quadruple...

Bridge 1: the physics

current is due to charge moving dipole is consisted of charges …

This bridge is complex, we do not need to take care of it.

1.Bioelectromagnetic Source models

1.Bioelectric Source models

Bridge 2: performance and mathematics

if a charge/dipole produces the same potential (EEG) of the actual current

---> charge/dipole is an equivalent source model of the current

1.Bioelectric Source models

Equivalent charge model (current source density)

1.Bioelectric Source models

Equivalent dipole model(Intracellular current)

By using Gauss Theorem

1.Bioelectric Source models

Equivalent “potential” model

1.Bioelectric Source models

(A) Equivalent charge model

extraqcellular current (EEG)

Neurophysiology of the equivalent model

a negative current source density (sink-negative charge )

a positive current source density(source-positive charge).

1.Bioelectric Source models

(B) Equivalent dipole model

A paired “negative charge- sink” and “positive charge- source”---> a dipole model

1.Bioelectric Source models

(1) Extracellular Current must flow in a regular way -- enough S/N -- -- to be recoded on the scalp surface

(2) equivalent Source model - Macroscale collectively activities-not the microscale intracellular current

Equivalent source model in practice

Comparison of the three models

Dipole Charge Potential Introduced year inEEG imaging

M Scherg ? 1985 D Yao ? 1996 Menendez etal 2000

parameter 3 1 1Current density Current source density Drive of current

Scalar or vector vector scalar scalar

Summary of Source models

three kinds of Source models each of them is an equivalent representation of the actual neuron “assembly”

2. 2D Imaging of brain activities

1) Image processing - Laplacian (deblurring the skull smearing effect)

2) Electric field analysis Cortical potential reconstruction Layer stripping(Equivalent dipole layer) Layer replacing(Equivalent charge layer)

Two approaches

2.2D Imaging of brain activities

h-radius of scalp, c-radius of the head

current source density(CSD) For a spherical head model (Yao, 2002)

Laplacian --- try to find the current emerge or disappear in the scalp layer

2.2D Imaging of brain activities

（ 1 ） Laplacian

（ 2) Electric field analysis

2.2D Imaging of brain activities

1.Cortical potential reconstruction (Sidman et al 1989;...)

2.source potential in infinite medium(Yao 2001)

3.Layer stripping(Equivalent dipole layer) (Freeman 1980, He Yao etal 2002)

4.Layer replacing(Equivalent charge layer)(Yao 2003)

（ 2) Electric field analysis

2.2D Imaging of brain activities

The characteristics of the spatial spectra of the above four imaging approaches

Equivalent charge layer approach -compared with Equivalent dipole layer (Yao 2003)

2.2D Imaging of brain activities

Forward(Three dipoles)

2.2D Imaging of brain activities

Inverse

2.2D Imaging of brain activities

Forward (four charges)

2.2D Imaging of brain activities

Inverse

2.2D Imaging of brain activities

Application

The source models may be:

Dipole -- Potential -- charge

3.3D Imaging of brain activities

VEPsVEPs

ECEC

ED AED A

ED XED X

ED YED Y

ED ZED Z

Real ERP result

1) Charge Loreta （ He,Yao and Lian, IEEE TBME, 2002 )

Charge Vs Dipole model: lower computation complexity, and may image both charges and dipoles

3.3D Imaging of brain activities

2) A Self-Coherence Enhancement Algorithm （ Yao et al 2001)

3.3D Imaging of brain activities

1) A Self-Coherence Enhancement Algorithm （ Yao et al 2001)

3.3D Imaging of brain activities

Step 1

Left:

Actual sourceRight:

LORETA

1) A Self-Coherence Enhancement Algorithm （ Yao et al 2001)

3.3D Imaging of brain activities

Two unknown parameters: K and alfa

Step 2 V G X

1) A Self-Coherence Enhancement Algorithm （ Yao et al 2001)

3.3D Imaging of brain activities

Comparing the NBIs of the solution and the actual source to chose a proper K

Actual neuronal source distribution is of neurophysiological smoothness. By defining a NBI (normalized blurring index )

Step 3

Determine alfa

Determine K

1) A Self-Coherence Enhancement Algorithm （ Yao et al 2001)

3.3D Imaging of brain activities

Step 4

Reference is the oldest problem of EEG

There is not a point that its potential is zero all the time (Geselowitz, 1998 ）

A unitary reference is the best and ideal case

4. EEG Reference problem

EEG recordings

4. EEG Reference problem

（ Yao, Physiol Meas, 2001 ）

Temporal waveform

Real signal Average REST

Method:Average ref:Va=GaX VaGaX

Inf ref V=GX VaGGaVaGaGGXV )()(

4. EEG Reference problem

Change of Spectra

Real signal Average REST

The reference may have a large effect on the spectra

EEG/ ERP Lab at UESTC

Thanks