osl-dynamics, a toolbox for modeling fast dynamic brain activity

Area of Science:

• Neuroscience
• Computational Neuroscience
• Machine Learning

Background:

• Neural activity exhibits complex spatiotemporal structures crucial for cognition.
• Modeling fast (tens of milliseconds) and transient brain dynamics presents significant methodological challenges.
• The precise timing of cognitive events is often unknown a priori.

Purpose of the Study:

• To present the OHBA Software Library Dynamics Toolbox (osl-dynamics), a Python package for analyzing neural dynamics.
• To enable the identification and description of recurrent brain activity on rapid timescales.
• To provide novel summary measures for understanding cognition, behavior, and disease.

Main Methods:

• Development of a Python-based software library, osl-dynamics.
• Utilizing machine learning generative models adaptable to neuroimaging data.
• Applying models to analyze spatiotemporal and spectral characteristics of brain activity with minimal assumptions.

Main Results:

• osl-dynamics can identify and characterize brain dynamics on timescales as fast as tens of milliseconds.
• The toolbox integrates with various neuroimaging data types (MEG, EEG, fMRI, LFP, ECoG).
• Novel summary measures of brain dynamics are provided for enhanced analysis.

Conclusions:

• osl-dynamics facilitates the modeling of fast dynamic processes in the brain.
• The toolbox enhances the study of brain function, cognition, behavior, and neurological disorders.
• It offers a powerful approach to uncovering the rich spatiotemporal structure of neural activity.