Over the years, we have developed several computational techniques for the statistical analysis of sets of electron micrographs of biological macromolecules. These methods include various types of factorial analysis (correspondence analysis, principal components), outlier detection schemes, statistical criteria for the quantitative assessment of spatial resolution (spectral signal-to-noise ratios), and a fast algorithm for the determination of the scaling factors between two micrographs. A current goal is to be able to combine a larger number of micrographs with slight disparities in magnification, in order to obtain higher-resolution 3-D reconstructions of icosahedral viruses. For this purpose, we are developing new spline-based algorithms for various geometric corrections (arbitrary scaling, translation, and rotation). These methods are based on a general least squares principle.
