research.neuroinformatics | |||
The current efforts of our neuroinformatics section are building upon a variety of research projects and resources. These can be broken down into a number of categories: |
|||
Databasing | |||
As
the literature on neuroscientific phenomena grows, so does the immensity
of the datasets which support it. Efficient storage and retrieval of these
data has accordingly become an increasingly important task in this field.
Of equal value, perhaps, is the capability to integrate and query these
datasets in novel ways, and thereby obtain interesting new insights. One
of our chief assets is a database collating the existing literature describing
anatomical tracing studies performed on macaque monkeys. This database
is called, fittingly, Collations of Connectivity data on the Macaque brain,
or CoCoMac (www.cocomac.org).
CoCoMac, which is publicly available, is being utilized both to query
connectivity-related information on the macaque brain, and as a constraint
upon neuronal network models of both macaque and, to some extent, human
brains. |
|||
Structural Relationships Structural connectivity is the key to understanding how the brain functions. Often, however, this understanding is impeded by confusion over the relationship of structures referenced by separate studies (top). By combining the data obtained in these studies, which are typically described using distinct parcellation schemes, we can use CoCoMac to integrate this knowledge by comparing the logical and geometric relationships between them (bottom). |
|||
Structural and Functional Connectivity By generating adjacency matrices like the one pictured here, we can use CoCoMac to analyze structural and functional connectivity patterns for properties such as centrality, clustering, flow, reachability, etc. Such characteristics give us some insight into the brain's organization and generate novel hypotheses for further experimentation or modelling. These matrices can also be used as constraints on forward modelling techniques which seek to simulate the dynamic functionality of neuronal networks on various scales of space and time. |
|||
Visualization | |||
Visualization
of neuroscientific data is important for a number of reasons. For purposes
of presentation, a well-designed visualization has the capability to provide
an intuitive illustration of modelled phenomena where words can fall short.
It is also quite useful to the researcher to have a visual means of interpreting
his or her model and the data obtained from it. At present, we have developed
Paxinos3D, a visualization tool portraying a rhesus monkey brain atlas
(Paxinos
et al., 2003) in 3D space. Further such tools are in the works. |
|||
Network Analysis | |||
An
interesting question, given a large-scale connected brain network model,
is: what structural and functional patterns can be extracted from this
model? This is useful, for instance, for understanding the segmentation
of the brain into structural and functional clusters, which may subserve
specialized functions. Such information can also be used to analyze the
effects of localized lesions upon these connectivity patterns. Ultimately,
it would be desirable to generate functional predictions from these analyses,
and data obtained from real subjects can be used to test and refine these
predictions. Our group is currently working on a number of approaches
to address such questions. |
|||
Computational Modelling | |||
Given
a ever-expanding mass of noninvasive structural and functional imaging
data obtained from humans, along with an increasing institutional capacity
for massive computations, large-scale computational modelling is perhaps
one of the most promising frontiers of modern neuroscience. Our group
is currently exploring mean field modelling (MFM) of cortical activity,
combining anatomical geometry information extracted from T1-weighted imaging,
structural connectivity information from CoCoMac and diffusion-weighted
imaging, and functional information obtained from EEG, MEG, and fMRI techniques. |
|||