最近Natureに掲載されたGeometric constraints on human brain
functionなど脳のダイナミクスの解析法を中心に活躍されている、Ben D.
Fulcherさんに広島大学とzoomのハイブリッドでの講演をしていただく予定です。
概要:Like many systems in the world around us, the brain’s is a physical system with
complex
activity patterns that evolve through time and can be measured in the form of
multivariate
time series. We now have unprecedented data on brain structure, including
gene-expression
atlas data with high spatial resolution and whole-brain coverage, as well as
intricate
recordings of the brain’s activity dynamics. What representations of the brain allow
us to
find informative patterns in these data that clarify how the brain works in health
and
disease? In this talk I will introduce different ways of treating the brain as a
complex
dynamical system, including a discrete network representation (a connectome) and a
physical
representation (in terms of spatially embedded gradients and distributed modes). I
will also
provide an overview of related methods that we have developed for quantifying brain
dynamics, including time-series patterns of specific brain areas, as well as
pairwise, and
distributed coupling patterns (implemented in our hctsa and pyspi
software
packages). I will make reference to some specific recent applications, including
inferring
biomarkers of psychiatric disease, extracting data-driven representations of sleep
dynamics,
quantifying the effects of brain stimulation, and characterizing resting-state EEG
and fMRI
data.
Key Refs:
- Fulcher & Jones (2017). hctsa: A computational framework for
automated
time-series phenotyping using massive feature extraction. Cell Systems.
link
- Cliff et al. (2022). Unifying Pairwise Interactions in Complex Dynamics.
arXiv.
link
- Fulcher et al. (2019). Multimodal gradients across mouse cortex. PNAS.
link
- Pang et al. (2023). Geometric constraints on human brain function.
Nature. link
第2回Computational Neurology Club AI summary