The Connectome: Automated Submicron Reconstruction of Brain Circuitry

Papers describing early results from connectome projects are beginning to be published (see my earlier blog post “Mapping the Brain’s Connections: the Connectome“).

A new paper focused on automating the full three-dimensional reconstruction of brain tissue titled “Ultrastructural Analysis of Hippocampal Neuropil from the Connectomics Perspective” was published September 23, 2010 in Neuron.

This paper describes how researchers densely reconstructed four volumes of rat hippocampus neuropil using serial section transmission electron microscopy. By using both manual and automated techniques they were able to assess the feasibility of automating reconstructions on a scale impractical for manual reconstructions.

In addition, the researchers assessed the reliability of methods typically used to predict synaptic connectivity. They demonstrated that these methods (various forms of what is known as Peters’ rule) fail to accurately predict synaptic density on dendrites. The authors found a strong correlation between the density of synapses and dendrite caliber and no correlation between the caliber and the density of available axons.

The ultimate goal is to automate reconstruction of complete vertebrate circuits at the resolution of synapses using data collected using serial section transmission electron microscopy. There remain major hurdles in data acquisition and analysis that must be overcome before the goal may be reached. Nevertheless, it will be achieved some day. Then we’ll be able to simulate the reconstructed circuits using data from signal processing research at the neuropil level (see my earlier blog post “Dendrites Provide High Fidelity Discrimination Between Different Spatial-Temporal Sequences“).

Other related blog posts:

Mapping the Brain’s Connections: the Connectome

Dendrites Provide High Fidelity Discrimination Between Different Spatial-Temporal Sequences