Comparing Functional Networks of the Brain:
An Introductory Tutorial using Python
Randy Heiland1 and Olaf Sporns2
The analysis of brain networks is a challenging endeavor, even for experienced
researchers. For younger students, the concepts, mathematical formulations,
data formats, and variety of software tools can be especially daunting.
In addition, the software may be commercial, making it less accessible; it may
only work on certain operating systems or only be provided as source code that
requires compilation; or it may have a user interface with a steep learning curve.
The BioVis 2014 Data Contest, comparing resting state functional connectivity networks
of the human brain, provides an opportunity to demonstrate the freely
available, open source Python scripting language for such analyses. We provide
an easy to follow tutorial, with simple scripts and explanatory documentation,
that will hopefully encourage students to take an active, hands-on approach
to understanding how brain networks can be analyzed and visualized. Using a
Python distribution that includes many useful scientific packages, we begin by
visualizing various anatomical representations of the brain, in different data formats.
Next, we show how the functional network datasets can be visualized as
both color-coded connection matrices and graphs. Finally, we demonstrate how
one might approach the challenging problem of comparing pairs of networks.
To address the primary challenge of the data contest, we discuss and propose
a mapping from 30 unknown test subjects to 18 known contest subjects. Our
primary goal is to help students feel more empowered to perform brain network
analysis, as well as other, more general, data analysis projects.
1 Pervasive Technology Institute, Indiana University
2 Department of Psychological and Brain Sciences, Indiana University
Presentation
Modularity structures for "known" subjects (click for larger image)
