Prof. Ulla Ruotsalainen: Statistical MAP-EM reconstruction methods for
tomography imaging with missing data
Optimally, tomography imaging is based on complete profile measurements from all
angles. Image formation is then an inversion problem that tries to solve from
which kind of cross-sectional distribution the profile measurements originate.
However often some measurement angles are missing because of the technical
structure of the scanner or metal artefacts that change the information in the
profile measurements. In addition, measurements are often noisy. Our approach to
these problems has been development of statistical image reconstruction methods
that penalize noise in small local neighborhood applying Median Root Prior as
the prior information of the data. The developed MAP-EM reconstruction methods
have been successful in producing good quality images in positron emission
tomography (PET), single photon emission tomography (SPECT) and electron
tomography (ET). In my presentation, I will show the principle of these
reconstructions as well as the reconstruction results with simulated data and
with real measurements.
Prof. Lasse Lensu: Spectral Retinal Imaging and Image Analysis
Eye diseases have become one of the rapidly increasing health threats worldwide.
To diagnose these diseases cost-efficiently based on the first abnormal signs in
the retina, it is necessary to develop new imaging techniques to complement the
standard red-free and colour retinal imaging. Spectral retinal imaging increases
the colour resolution from the common greyscale and three-channel
representations to a new one with six or even tens of channels depending on the
system. The increased colour resolution has been shown to enable new type of
analysis and it has been applied, for example, to measure the oxygen saturation
in the retinal blood vessels, estimate the concentrations of retinal molecules
and detect lesions associated with the eye diseases.
Lasse Lensu is a Professor of Computer Science and Engineering at Lappeenranta
University of Technology (LUT), Finland. He received his D.Sc. (Tech.) degree in
Computer Science and Engineering in 2002 from the Department of Information
Technology of LUT. His research interests include digital imaging and image
processing applications, computational vision and medical image analysis. Prof.
Lensu is the head of the Machine Vision and Pattern Recognition Laboratory of
LUT, and he has contributed to the technology transfer to two spin-off companies
from the university.