On the exploitation of a commercially available parallel processing architecture for medical imaging

A strategy for assessing the suitability of a general-purpose parallel processor computer architecture for medical imaging is presented. The research objectives and planned phases of the research are outlined. The results of a survey of previous work in this field are summarized. A methodology for assessing the performance of multiprocessor medical imaging machines is introduced. An algorithm developed to threshold voxel-based image data on a hypercube is described. The current status of the research and future activities are presented.<<ETX>>