The Computational Core is the programmatic structure through which the model-based image reconstruction Program theme will be exploited. It will be responsible for both a service function and a development role with the Program. Economy of effort will be realized through (i) centralized computerized systems and software management, (ii) common uses of software interfaces to data acquisition, hardware through PC-bus controllers and (iii) development and maintenance of computational libraries and tools related to model-based image reconstruction methods. In addition to the economy of effort that the Core provides, it also presents opportunities to exploit the model-based image reconstruction theme across the research Projects in a way which enhances cross- fertilization of ideas and methods. Specific developmental activities of the Core in this regard we to (1) develop and implement parallel processing paradigms for model-based image reconstruction, (2) develop and implement three-dimensional dual meshing, and (3) develop and implement adaptive dual meshing, first in two and then in three dimensions and (4) investigate and implement new concepts in model- based image reconstruction including multi-component objective function minimization and multi-spectral imaging strategies as driven by Project interests. In addition, a significant new computational resource- a shared-memory multi-processor workstation-has been proposed with a 50% cost-sharing arrangement. This system would not only speedup computations by a factor approximately equal to the number of available processors but it would also provide access to considerably larger amounts of memory than typically resident on a single processor platform. The substantially increased memory also serves to significantly increase the discretization scale than can be achieve which is critical for realizing maximal image resolution with model-based methods. The Computational Core will serve the four Project components on an equal basis while also interfacing with the Clinical Core on critical data management and analysis issues pertaining to the clinical breast imaging exams.
Showing the most recent 10 out of 129 publications
