Bayesian Inference for Functional Dynamics Exploring in fMRI Data.

Area of Science:

• Neuroscience
• Computational Biology
• Statistical Modeling

Background:

• Functional magnetic resonance imaging (fMRI) generates complex time series data.
• Analyzing functional interactions and temporal segments in fMRI is challenging.
• Bayesian inference offers a robust framework for modeling uncertainty and dependencies.

Purpose of the Study:

• To review state-of-the-art Bayesian inference methods for fMRI data analysis.
• To address the challenge of exploring functional interactions and temporal boundaries in fMRI time series.
• To introduce and compare popular Bayesian models for fMRI data.

Main Methods:

• Review of Bayesian inference techniques applied to fMRI.
• Focus on methods for detecting change points in fMRI data (magnitude and connectivity).
• Comparison of Bayesian Magnitude Change Point Model (BMCPM), Bayesian Connectivity Change Point Model (BCCPM), and Dynamic Bayesian Variable Partition Model (DBVPM).

Main Results:

• Bayesian inference effectively encodes dependence relationships and uncertainty in fMRI data.
• Identified methods can detect functional connectivity change points and infer interaction patterns.
• Comparison highlights the applications and strengths of BMCPM, BCCPM, and DBVPM.

Conclusions:

• Bayesian inference is a powerful tool for computational modeling of fMRI data.
• Emerging Bayesian models are crucial for advancing the understanding of brain functions.
• These methods enhance the analysis of temporal dynamics and functional connectivity in the brain.