Overview
Get the Flash Player to see this player.
"From Images to Brain Function"
Neurons are the basic computational subunits of the brain. Being single cells, they are of small physical dimension and thus difficult to study. Typical cell body diameters are approximately 10 μm, with dendritic branches on the order of 1 micron or less. However, their small stature is in strong contrast to their computational power. Neurons are complex electrochemical compartments capable of integrating hundreds of thousands of input signals on the millisecond time scale.
In spite of the technical challenges, fully elucidating single neuron function is crucial to understanding the brain.
Motivation
One can reasonably argue that the brain is the most complex system known to man. The mammalian brain is composed of 100 billion cells, with up to 100,000 intracellular connections between each of those cells. This high interconnectivity supports complex interactions between each of these subunits to form sophisticated processing networks. The ensemble network activity gives rise to the familiar behavioral results of the higher organisms: sensory responses, motor activity, and, in humans, cognition. Further underscoring the great need for increasing our knowledge of brain function is the potential for medical advances: successfully treating cognitive dysfunction, injury, and neural diseases would have a substantial societal impact.The very complexity that allows these high-level capabilities also makes the brain difficult to study and understand. It is therefore necessary to take a more focused approach. The basic subunit of the brain is the single cell, called a "neuron." Neurons are unique from other cells in that they possess branching processes called "dendrites," conduct electrochemical impulses, and have dynamic sensitivity.
It is also important to note that neurons are extremely heterogeneous. Throughout the nervous system, one observes a wide range of dendritic morphologies (Figure 1) in addition to a wide range of electrical behavior. This heterogenity in both features suggests a structure-function relationship.
Goal
The goal of this project is to develop a computational and experimental framework to allow real-time mapping of functional imaging data (e.g., spatio-temporal patterns of dendritic voltages or intracellularions) to neuronal structure, during the very limited duration of an acute experiment.Objectives
- To develop the theoretical and computational framework for accurate and robust morphological reconstruction of living, fluorescently-labeled dendrites from optical serial sectioning data.
- To constrain a compartmental neuron model with data from the reconstructed morphology.
- To use the simulation results in combination with the functional imaging data to increase our understanding of single-neuron computation.
The current state-of-the-art for translating structural images of neurons to cylinder models requires laborious manual dendrite tracing by a human. This tedious process takes many hours for typical morphology complexity and suffers from:
1. An inability to reconstruct live neurons under experimental investigation
2. Tissue fixation problems
3. Imaging resolution limitations
4. Operator subjectivity.
These drawbacks are problematic for two reasons. First, the reconstruction results--and thus the simulation results--are not available at the time of the experiment, eliminating the possibility of tight integration of experiments and simulations. Second, the accuracy of simulation results is dependent on the fidelity of the reconstructed morphology to the real neuron: the physical dimensions of the neuron dictate the cable properties assigned to the model.
We address these problems by avoiding them altogether. Our software suite generates cylinder models from 3D fluorescence images. This has the advantages that the output cylinder model is rapidly-done, accurate, and robust.
1. Rapid: minimal to no human intervention; doesn't require tissue fixation
2. Accurate: algorithms are objective, relative to human subjectivity; uses high-resolution fluorescence data
3. Robust: algorithms are objective, relative to human subjectivity
With the accurate morphological model in hand, we can integrate multi-site functional imaging results and model simulation results to increase our understanding of single neuron function.
