Three-dimensional distortion correction applied to tomographic reconstructions of sectioned crystals.

Unaveraged tomograms of embedded and sectioned insect flight muscle show not only structural variation within the individual unit cells but also defects in the filament lattice such as lack of straightness of the filaments and wrinkling of the sections. Adaptation of conventional 2D crystal image processing methods usually average out these structural variations and lattice defects thereby reducing the resolution and information content of the 3D reconstructions. We present here a computational method to correct the reconstructed 3D object for these distortions to make it suitable for further analysis. The correction is an extension of conventional 2D cross-correlation techniques applied to a 3D image. These procedures determine the unit cell positions of the distorted crystal and use a spline-fitting approach to calculate a mapping function from the observed positions that is subsequently used to reinterpolate the image onto a regular grid. The resulting reinterpolated image is corrected for both filament straightening as well as section wrinkling.