Neuroanatomical algorithms for dendritic modelling.

Abstract

The complexity and variability of dendritic morphology constitutes a fascinating challenge to the investigation of the structure-activity-function relationship in the nervous system. Computational modelling has recently emerged as a powerful approach for the quantitative anatomical characterization of dendrites. The key idea is to design a stochastic algorithm to generate digital structures that are statistically indistinguishable from those of real neurons of a given morphological class. The set of parameters used by this algorithm would then constitute a complete and accurate description of that morphological class. We review the strengths and weaknesses of the major types of algorithms used to model dendrogram properties, including those based on branch diameter and on distance from the soma. We also describe some approaches to the simulation of dendritic orientation and three-dimensional geometry. Finally, we discuss the environmental influences on dendritic morphology (especially the presence of axons, other neurons, and anatomical boundaries) and thus the need to include models of the tissue volume in the algorithmic description of dendrites.