You are on page 1of 2

BMEn 5401

Project #1
DUE Nov 9, 2017

The forward problem is essential to model-based inverse imaging techniques. In


bioelectromagnetic source imaging, an important forward problem is to solve the electric
/ magnetic field generated by equivalent dipole sources. The bioelectromagnetic source
imaging is to solve the inverse problem of estimating source distribution from
electromagnetic measurements.

In the figure on
the right, there are two
layers. The upper layer is
the sensor layer, where
NxN electrodes are
uniformly distributed
over an x-y plane. (The
coordinate system is
defined by origin o, three
orthogonal directions x, y,
and z as indicated in the
figure.). The bottom layer
is the source layer which
is also in an x-y plane and
parallel to the sensor
layer. The MxM possible
source locations (regarded as a distributed source model) are indicated by small circles
and there is no constraint for source orientations. The z axis passes through the centers
for both layers.

A. Forward Problem:
Assuming the top layer is the body surface and the bottom layer the current dipole
source distribution. Consider the volume conductor as an infinite homogeneous medium.
Use equivalent current dipole models to simulate the cardiac sources. Use 1, 2, or K (K>2)
dipole sources in the source space and perform the forward computation.
1. If this forward modeling is for cardiac application, assuming only 1 or 2 equivalent
dipole sources are located within the heart, compute the electrical potential and
magnetic field (maybe one component) on the sensor surface (top layer) and plot
them. Display the distributions of electrical potential and magnetic field (you may
consider one component of the magnetic field) corresponding to various parameters.
Use realistic physical units. Discuss the effect of the following parameters on the
electrical potential and magnetic field: a) inter-dipole distance; b) measurement noise.
2. If this forward modeling is for brain application, consider K = 1, 2, K (K>2) and
repeat other procedures as in step 1) above.

1
3. Write a report including equations, results plots, discussion, and computer codes you
used. Write a paragraph on the difference between cardiac and brain modeling from
this simple example. Write another paragraph to discuss the features of electrical and
magnetic measurements.

B. Inverse Imaging Problem:


In the source-sensor configuration, without considering the unit of values for all
parameters, assuming N=10, M=8, h=4, inter-sensor distance b=1, inter-source distance
a=1, and conductivity σ=0.33 S/m (infinite homogeneous medium), complete the
following problems. You may choose to use either electrical or magnetic field which you
obtained in the problem A.
1. Write codes to generate a transfer matrix which relates the source space to sensor
space. What is the size of the transfer matrix? (Hint: Don’t forget to take source
orientations into consideration.) Assuming there are two dipole sources at location 1
and location 2 (solid black circles), use the generated transfer matrix to obtain the
distribution of electrical or magnetic field over the sensor layer. The sources moments
are (-0.2, -0.2, 1) for dipole 1 and (0.5, 0.5, 0.1) for dipole 2. Show the distribution of
the electrical or magnetic field over the sensor space.
2. Solve the inverse imaging problem based on the obtained field distribution in step 1
without additive noise. Show your results.
3. Solve the same problem as in step 2 with 10% additive noise. Show your results.
(Hint: The inverse operators are available in many forms. Truncated SVD
regularization is suggested as a simple one).
4. Increase the density of possible source space from M=8 to M=12 uniformly
distributed sources without changing the size of source layer. It means to decease the
inter-source distance. Repeat the above work and show your results.
5. Discuss whether the M=8 case is an over-determined system or underdetermined
system. What about the M=12 case? What is the difference between these two cases?
6. Write a report including equations, results, codes, and a discussion of the solutions,
results and interpretation (consider applications to cardiac imaging and brain
imaging).

You might also like