I have had students work on image formation projects
in PET, SPECT, X-ray CT, MRI, microscopy, radiation therapy,
and biomedical signal detection.
It is not feasible for me to keep this list current.
To get a sense of what my group is doing,
look under the "publications" section
of my web page for papers that are in review and in press,
and for recent conference papers.
To see my current projects,
look at my
vita (cv),
which will be more current.
Below are some past projects,
from back in the days when I had fewer projects
so I had more time to update this...
Statistical Methods for Attenuation Correction in ECT
NIH R29 CA 06711
Student: Hakan Erdogan
One of the most important physical effects
in positron emission tomography (PET)
is photon attenuation.
To correct for the affects of attenuation,
PET scans are preceded by a transmission scan,
which is usually very noisy.
Traditional attenuation correction methods
are suboptimal
since they do not fully account
for the noise in the transmission measurements.
We are developing methods for estimating
the attenuation correction factors
based on statistical models
for the transmission measurements.
This work has led to new algorithms
for tomographic image reconstruction.
We are also developing methods
for joint estimation of transmission and emission data
from "post-injection" transmission scans,
which are contaminated by emission counts.
Spatial Resolution Properties
of Penalized-Likelihood Image Reconstruction Methods
Whitaker Foundation
Student: Web Stayman
We have shown that traditional methods
for regularizing ill-conditioned problems
such as tomographic image reconstruction
lead to shift-variant spatial resolution
when the measurement statistics
are nonstationary.
In this project we are developing
improved regularization methods
that yield uniform spatial resolution
even for shift-variant tomographic systems.
Estimation Strategies for Nuclear Medical Imaging
PI: W. Leslie Rogers (with A. O. Hero and N. H. Clinthorne)
NIH R01 CA 54362
Student: Mehmet Yavuz
In this project we are developing
improved image reconstruction methods
for imaging small animals
with a Michigan-designed SPECT
(single-photon emission computed tomography)
system known as SPRINT.
The project includes statistical methods
for analyzing and predicting system performance
based on lower bounds on mean-square error.
Positron Emission Tomography of Breast Carcinoma
PI: Richard L. Wahl
NIH R01 CA 52880
In this project
we are evaluating the efficacy of fully 3D PET
for imaging breast cancer.