The broad goal of the proposed work is to implement and extend a new method for electron-microscopic autoradiographic (EMA) analysis so that it can be made widely available to biomedical investigators. The new method has been shown in simulation and real-data general use. The new method is based on the maximum-likelihood method of statistics and uses the expectation-maximization algorithm, which leads to recursive equation that was, in its original formulation, too computationally demanding to be practical for conventional computers. Recent reformulations of the algorithm render it computationally practical on small PC-class computers generally available to laboratory investigators.
Specific aims are: (1) to implement the new method for small computers in a generalized form that can be accommodated by a wide range of commonly available machines, and offer it to the national community; (2) to develop an EMA simulator which uses the experimental data to produce with every analysis of sets of micrographs, a statistical measure of the reliability of the estimates produced, and which functions as a utility for establishing the experimental requirements necessary for testing biological hypotheses; (3) to determine the fundamental resolving power of the method and how it is influenced by factors affecting point-spread for different radio-isotopes; and (4) to develop an automated EM-image segmentation (structure delineation) method in order to reduce the tedium inherent in current manual methods for data acquisition. The significance of the new method is that in comparison to the cross-fire method (1) it yields estimates of tracer concentrations with much greater accuracy, (2) it is able to produce biologically meaningful results from far fewer EM images and/or at much lower levels of radioactivity, (3) it guarantees global convergence on the maximum-likelihood estimate, thereby eliminating the need for ad hoc """"""""rim compartments."""""""" The power of the new method has been demonstrated in a study of sarcolemmal uptake of arachidonic acid by injured myocytes, which has yielded new insights into the pathogenesis of ischemic myocardial injury.
| Joshi, S; Miller, M I (1993) Maximum a posteriori estimation with Good's roughness for three-dimensional optical-sectioning microscopy. J Opt Soc Am A 10:1078-85 |
| Roysam, B; Maffitt, D R; Miller, M I et al. (1992) A personal computer based implementation of the maximum-likelihood method of analysis of electron microscope autoradiographs. Microsc Res Tech 20:73-86 |
