Explore the landscape of brain dynamics during ongoing cognition using Mapper in this 29-minute talk by Samir Chowdhury. Delve into modern non-invasive neuroimaging methods like fMRI and their high-resolution spatial patterns of brain activity. Learn about the Mapper framework's application in revealing whole-brain activity organization for multitask fMRI data, accounting for nonlinear and autocorrelation structures in neuroimaging data. Discover a quantitative approach using distance distributions to predict task performance at a granular level. Examine the talk's coverage of MRI analysis, precision psychiatry trends, the Mapper pipeline, continuous multitask paradigm, nonlinear distance assessments, parameter selection heuristics, and validation through core-periphery and community structure analysis. Gain insights into postdoc analysis via distributions and optimal transport distances for quantifying task transitions. Conclude with an exploration of future directions in this field of neuroimaging and data analysis.
Exploring Brain Dynamics During Ongoing Cognition Using Mapper
Applied Algebraic Topology Network via YouTube
Overview
Syllabus
Intro
Overview of MRI analysis
Current trends and needs in precision psychiatry
The Mapper pipeline
Experiment design: the continuous multitask paradigm
Simple access to nonlinear distances: the kNN graph and Isomap
Assessing the discriminative power of linear vs nonlinear distances: distance distributions
A heuristic for parameter selection
Validation: core-periphery and community structure relate to behavior
Posthoc analysis via distributions
Optimal transport distances quantify transitions between tasks
Future directions
Taught by
Applied Algebraic Topology Network
