Third: most intriguingly, we found that the scaffold's structure dictates its information flow. The borders of loops handle redundant information, while the connections spanning these loops exhibit high synergy. It's a beautiful metaphor: integration happens within the voids. 8/n

Second, while FC fingerprints often draw from features confined to particular brain networks, scaffolds gain their power from inter-network connections. In other words, identity seems to live in how different large-scale brain systems interact 7/n

First, scaffold-based fingerprints achieved near-perfect identification accuracy (~100%), outperforming FC-based methods (~90%). Even more 🤯, they remained robust across different preprocessing strategies, brain atlases, and dramatically shortened scan times. 6/n

To better capture this richer structure, we use homological scaffolds. Picture incrementally building a brain network by adding connections in order of strength. As we do this, loops begin to form. The scaffold is made of all edges that participate in these mesoscale loops 4/n

We show that the topological structure of brain connectivity—, homological scaffolds—offers a powerful and highly robust way to identify individuals based on their functional brain organization. 2/n

Thrilled to finally share a huge piece of my PhD project!
Our paper “The Topological Architecture of Brain Identity” is now out on bioRxiv: biorxiv.org/content/10.1... 🧵 1/n