Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Stopping rules, Bayesian reconstructions and sieves

Conference ·
OSTI ID:5520876
During the last two years, our group has been working on the concept of feasible'' images in the context of image reconstruction for emission tomography (ET). The concept has been discussed previously in image processing in astronomy, but had not been applied to tomographic image reconstruction until we related the deterioration of Maximum Likelihood Estimator (MLE) images at large number of iterations to the unfeasibility of the resulting images. We have described a feasible image as an image that, if it were a true radioisotope distribution in a patient, could have generated the measured data by the Poisson process that governs the radioactive decay process. Formal definitions of feasibility, as well as tests that can be applied to computer generated data or to real tomographic data, have been given in the literature cited above. 18 refs., 3 figs.
Research Organization:
Lawrence Berkeley Lab., CA (USA)
Sponsoring Organization:
DOE/ER
DOE Contract Number:
AC03-76SF00098
OSTI ID:
5520876
Report Number(s):
LBL-26499; CONF-890605--2; ON: DE90001420
Country of Publication:
United States
Language:
English