Robust Recovery of Temporal Overlap Between Network Activity Using Transient-Informed Spatio-Temporal Regression

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Systems Neuroscience

Background:

• Functional magnetic resonance imaging (fMRI) is a key tool for understanding brain function.
• Analyzing dynamic brain activity requires advanced methods to capture temporal patterns.
• The innovation-driven co-activation pattern (iCAP) approach identifies transient brain activity but faces challenges in temporal profile recovery.

Purpose of the Study:

• To develop a novel back-projection method for fMRI data analysis.
• To improve the accurate recovery of temporal profiles for spatially and temporally overlapping brain networks.
• To overcome limitations in current iCAP methods caused by spatial dependencies.

Main Methods:

• Proposed a novel back-projection method integrating spatial and temporal constraints.
• Validated the method using simulated fMRI data.
• Evaluated the approach on experimental fMRI data to assess its performance in preventing overfitting and underfitting.

Main Results:

• The new back-projection method significantly improves the quality of fitted time courses for iCAPs.
• Optimized spatio-temporal constraints effectively prevent overfitting and underfitting in experimental data.
• The method successfully recovers spatial and temporal properties of large-scale brain networks.

Conclusions:

• The developed back-projection method enhances the analysis of dynamic brain activity using fMRI.
• This approach allows for a more accurate characterization of brain activity as recurrent co-activation and co-deactivation of networks.
• Opens new possibilities for exploring the brain's dynamic core and large-scale network interactions.