Predicting functional connectivity from structural connectivity via computational models using MRI: an extensive...

Area of Science:

• Neuroscience
• Computational modeling
• Brain connectivity

Background:

• Magnetic resonance imaging (MRI) studies structural connectivity (SC) and functional connectivity (FC).
• MRI limitations necessitate computational models to infer underlying physiological mechanisms.
• The predictive power of these models remains inadequately understood.

Purpose of the Study:

• To compare the predictive performance of seven computational models of brain connectivity.
• To assess how well simulated functional connectivity (FC) predicts empirical FC.
• To analyze model and data relationships using graph theory across spatial scales.

Main Methods:

• Compared seven computational models predicting FC from SC.
• Evaluated prediction accuracy of simulated vs. empirical FC.
• Applied graph theory analysis to simulated and empirical FC data at three spatial scales.

Main Results:

• Model predictive power varied significantly with spatial scale.
• The simplest model, simultaneous autoregressive (SAR), showed superior and consistent performance.
• Empirical FC exhibited weak correlation with simulated FC and was peripheral in graph analysis.

Conclusions:

• Differences among computational models had minimal impact on their FC predictive power.
• Model capacity to predict FC from SC is moderate.
• A simple linear process, as embodied by the SAR model, underlies FC prediction from SC.