At TDA-Brain, we develop topological learning methods: principled statistical and machine learning frameworks that incorporate topological representations directly into modeling pipelines. Our work integrates persistent homology, graph topology, and geometric representations with modern machine learning to analyze complex, high-dimensional neuroimaging data. We focus particularly on multimodal data—including structural MRI, functional MRI, diffusion imaging, and brain connectivity networks—where topology provides a natural language for describing brain organization across spatial and temporal scales.
By embedding topological features into statistical inference methods and machine learning models, we aim to uncover latent patterns of network reorganization and resilience. We apply these approaches to conditions such as epilepsy and post-stroke aphasia, seeking to:
Characterize alterations in brain network topology
Identify biomarkers for disease severity and recovery
Improve outcome prediction and patient stratification
Reveal mechanistic insights into brain plasticity and dysfunction
Through the integration of topology, statistics, and machine learning, TDA-Brain advances a new framework for understanding brain network disorders—one that captures the geometry and connectivity of the brain as a dynamic, multiscale system.
Website updated on 3/11/2026