Figure 6.3 Example of variance estimation, applied to the assessment of significance of a feature in a 3D difference map. Two reconstructions are compared: one (selected slices in row a) derived from hemocyanin molecules that are labeled at all four sites, and another (selected slices in row c) at three sites only. Rows b and d: 3D variance estimation computed for reconstructions a and c, respectively, following the procedure outlined in the flow diagram of figure 6.2. Row e: difference volume a—c. Row f: 3D t-test map, computed for a 99% confidence level, showing the 3D outline of the Fab molecule. From Boisset et al. (1993b), reproduced with permission of Elsevier.
selecting k projections at random with replacement. This means that in each such permutation of the original projection set, some of the projections are missing and some occur repeatedly. From the n reconstructions, a voxel-by-voxel variance o2 is calculated. The sought-for "correct" variance is then estimated as
Was this article helpful?