We address the problem of building three-dimensional  connectivity maps for neurons from sectional electron microscopy.  Sectional data consists of a stack of very high-resolution, two-dimensional images that are oriented to capture cross sections of elongated neuronal processes. High magnification serial microscopy images have the potential to expand the field of neurophysiological modeling by providing ground-truth neuroanatomical data. However, their complexity and vast size make them impractical for human interpretation. This project aims at building automatic and semi-automatic tools to assist researchers in analyzing such data.

We have developed a new method for constructing statistical representations of ensembles of similar shapes that uses particle systems to represent surfaces non parametrically and optimally sample surface point correspondences.   We used this method to generate models for two clinical datasets: normal vs. Autistic neurological development.  Hypothesis testing on these models using a non parametric permutation test of the Hotelling T-squared metric (including false-discovery-rate (FDR) correction) reveals significant group differences. Colormap indicates the magnitude and direction of the linear discriminant.