EEG

Question 1

What is spherical head model?

Answer 1

Approximate the shape and electromagnetic properties of head using a three concentric sphere model

For brain, skull and scalp

Assume homogenous conductivity in each sphere

The potential on the scalp can be computed analytically by solving quasi-static (frequency <1 kHz) approximation of maxwell equations

Compute magnetic field from
These

Question 2

What are boundary element model?

Answer 2

Use the subjects mri to construct the surface of scalp of brain

Represent the surface by triangular measures

The smaller the triangles are, the more precise the surface is - higher computational results

BEM is a numerical model

Question 3

What is Finite element head model?

Answer 3

Five compartment (scalp,skull,CSF, brain grey and white matter)

Many thousand elements in each with different conductivity

Also use Electrical impedance tomography

Sagittal cit through finite element

DTI data used for estimation of grey/white matter conductivity

Question 4

What are the MEG head model

Answer 4

1. Single sphere
   - analytic
   - model head as a sphere, radial sources are invisible
2. Corrected spherical model
   - analytic
   - correcting for head shape deviations from sphere
3. BEM
   - detailed numerical model of volume currents

Question 5

What is distributed source reconstruction?

Answer 5

1. From an MRI create a cortical mesh with e.g. 3000 vertices
2. Place a dipole perpendicular to cortical surface area each vertex
3. For each dipole, we only need to estimate the strength j
4. Fewer sensors than sources (bad). Constraint needed

Question 6

What is ECD used to estimate?

Answer 6

Early components of ERP which are highly localised

Useful for subcortical reconstruction

Question 7

What is distributed solutions used for?

Answer 7

Reconstructing later components of ERPs

These more cognitive components are often highly distributed throughout cortex

Question 8

What does spherical model ignore ?

Answer 8

Anisotropy of conductivity

Question 9

What do ECD methods have?

Answer 9

Few parameters but are non-linear

Question 10

What do distributed solutions have?

Answer 10

Linear

But due to large number of parameters require additional constraints

Question 11

What do beamforming approaches use?

Answer 11

Spatial filtering and scanning instead of trying to fit the scalp data