Computational and dynamic models in neuroimaging.

Area of Science:

• Computational Neuroscience
• Neuroimaging
• Systems Neuroscience

Background:

• Neuroimaging analysis has been significantly enhanced by computational neuroscience.
• A distinction exists between models of brain function and models of neuronal biophysics.

Purpose of the Study:

• To review the impact of computational neuroscience on neuroimaging.
• To illustrate the role of functional and biophysical models in analyzing neuroimaging data.

Main Methods:

• Focus on optimal control and decision theory for functional models.
• Emphasis on dynamic causal modeling and neural-mass models for biophysical modeling.
• Utilizing generative models for hypothesis embedding and model comparison for hypothesis testing.

Main Results:

• Functional models offer mechanistic accounts of neuronal computations and latent mental states.
• Biophysical models, particularly neural-mass models, advance the analysis of hemodynamic and electrophysiological time series.
• Generative models and model comparison are crucial for hypothesis testing in neuroimaging.

Conclusions:

• Computational neuroscience provides essential tools for advancing neuroimaging.
• Both functional and biophysical modeling approaches enrich the scope and depth of neuroimaging research.
• Generative models and rigorous model comparison are key to future trends in imaging neuroscience.