Skip to Main Content
Summary form only given. In the study of biological processes like virus maturation, experimental situations arise where the sample is a mixture of virus particles in which each particle is from one of a few classes of identical particle. In order to use cryo electron microscopy to compute a 3-D reconstruction of each class of particle, a pattern recognition problem must be solved. A model-based statistical approach using the maximum likelihood criteria in which the unknown class labels are treated as nuisance parameters is described. An expectation-maximization algorithm is used to solve the maximum likelihood problem where, in order to compute reconstructions at biologically interesting spatial resolutions, a high-performance computing implementation has been developed on a cluster computer.
Statistical Signal Processing, 2003 IEEE Workshop on
Date of Conference: 28 Sept.-1 Oct. 2003