We applied Mapper to a massive single-subject neuroimaging dataset to visualize the brain's networks.

Mapper is an algorithm from topological data analysis that uses dimensionality reduction techniques (we used tSNE) to project high-dimensional data into a lower-dimensional embedding and creates a network structure that represents the "shape" of the original data.

We found that we can visually represent the modular organization of known brain networks and use graph theory metrics to infer behavior or physiological state.

To view the main results, you can scroll through ProjectReport.ipynb (there's a lot of scrolling).