- Feed these fingerprints into a neural network to help distinguish ASD vs typically developing participants tested on a subset of 871 subjects (Autism Brain Imaging Dataset)
In this work, we:
- Build directed brain networks from resting-state fMRI (using a simple time-lagged correlation)
- Summarise each network with a set of topology-based fingerprints (called Betti curves) that capture global patterns like how the network connects and forms loops
👉 The key takeaway: these topology-based features seem to capture useful structure in directed brain networks and can complement more standard approaches for connectome-based classification
🧠 Can brain network “shape” help detect autism? Our new fMRI study