Atlas home || Conferences | Abstracts | about Atlas

VIRTUAL OBSERVATORY: "Plate Content Digitization, Archive Mining and Image Sequence Processing"
April 27-30, 2005
Auspices of COST Actoin 283: "Computational and information infrastructure in the Astronomical DataGrid"
Sofia, Bulgaria

Organizers
Milcho Tsvetkov, Bulgarian Academy of Sciences, Space Research Institute, and Valeri Golev, Sofia University, Department of Astronomy

View Abstracts
Conference Homepage

Spatial Segmentation of Image Sequences Based on their Time Activity
by
Nikolaos P. Galatsanos
Computer Science Department, University of Ioannina, Ioannina, Greece 45110

There are many applications in medical imaging where one is interested in finding the areas of the image that exhibit the same time activity. Such applications occur in positron and single photon emission imaging as well as in perfusion studies with magnetic resonance imaging (MRI). In this talk we will present Bayesian methodology based on clustering to solve this problem. At first the dimensionality of the pixel observations is reduced using a probabilistic principle component model along the spatial dimension of the data. Then, a multidimensional Gaussian mixture model with spatial constraints is used for clustering. Examples from MRI perfusion studies of the heart and the brain will be shown.

Date received: February 28, 2005

Copyright © 2005 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # capb-24.