In this talk we discuss new algorithms for the reconstruction of positron emission tomography images, based on the Shepp-Vardi statistical model. We also include interesting applications of abstract measure theory to understanding the numerical divergence of the classical EM iteration procedures, and to obtain partial answers to an open question of convergence of a generalized EM algorithm. This talk also deals with the vital relationships between mathematics and engineering which resulted in this work, and continues to fuel current work.

1996-1997 Mathematics in High Performance Computing