The principal aim of the research is to provide a method of reconstruction for tomography (SPECT, PET, TOFPET, X-CT) which represents a substantial improvement in terms of image quality and instrument design flexibility over the presently available techniques. The new proposed method is based on calculating accurately the response matrix of an imaging instrument, optimizing its design by mathematical analytic methods and using the final response matrix as the probability function needed for a maximum likelihood estimator (MLE) reconstruction algorithm. In addition, we propose to study processor architectures for the purpose of designing and building a prototype processor that can open the way to a future cost-effective implementation of the MLE algorithm. The MLE algorithm has been shown to provide substantial improvements in image quality in PET tomography. Unfortunately the long computation times and/or the large amount of storage needed for its implementation make immediate use and even research on the method impractical. By providing a well validated group of programs that calculate the probability functions for an imaging instrument including effects due to incidence angle and position, energy resolution and cross-talk, and by developing an effective method of image analysis, we expect to optimize the MLE imaging problem. The demonstrated favorable computing power-to-cost ratio of microprocessor cluster architectures and array processors can then be exploited to implement the MLE algorithm. The proposed research program could have a substantial impact in all areas of medical research and diagnosis which use tomography by providing higher sensitivity, spatial accuracy, contrast, resolution design flexibility and better utilization of emitted radiation. This is particularly more important now, with the development of new high resolution instruments.
