PARALLELIZED FORMULATION OF THE MAXIMUM LIKELIHOOD-EXPECTATION MAXIMIZATION ALGORITHM FOR FINE-GRAIN MESSAGE-PASSING ARCHITECTURES

Recent architectural and technological advances have led to the Feasib ility of a new class of massively parallel processing systems based on a fine-grain, message-passing computational model. These machines pro vide a new alternative for the development of fast, cost-efficient Max imum Likelihood-Expectation Maximization (ML-EM) algorithmic formulati ons. As an important first step in determining the potential performan ce benefits to be garnered from such formulations, we have developed a n ML-EM algorithm suitable for the high-communications, low-memory (HC LM) execution model supported by this new class of machines. Evaluatio n of this algorithm indicates a normalized least-square error comparab le to, or better than, that obtained via a sequential ray-driven ML-EM formulation and an effective speedup in execution time (as determined via discrete-event simulation of the Pica multiprocessor system curre ntly under development at the Georgia Institute of Technology) of well over two orders of magnitude compared to current ray-driven sequentia lML-EM formulations on high-end workstations. Thus, the HCLM algorithm ic formulation may provide ML-EM reconstructions within clinical time- frames.