Characterization of task-free and task-performance brain states via functional connectome patterns

Area of Science:

• Neuroscience
• Computational Neuroscience
• Medical Imaging

Background:

• Resting-state fMRI (R-fMRI) and task-based fMRI (T-fMRI) are standard tools for studying brain activity.
• Challenges exist in ensuring R-fMRI and T-fMRI accurately reflect task-free and task-performance states due to uncontrolled mental status and behavior.
• Distinguishing between these states is crucial for reliable interpretation of fMRI data.

Purpose of the Study:

• To develop a novel computational approach for characterizing and differentiating brain functional states (task-free vs. task-performance).
• To accurately represent and analyze whole-brain functional connectome patterns from fMRI data.
• To identify discrepancies between expected and actual brain states during fMRI scans.

Main Methods:

• Representing brain functional state using whole-brain quasi-stable connectome patterns (WQCP) from 358 cortical landmarks.
• Applying sparse representation to learn atomic connectome patterns (ACPs) for task-free and task-performance states.
• Comparing learned ACPs from R-fMRI and T-fMRI datasets and investigating outliers using functional activation detection.

Main Results:

• Learned ACPs for R-fMRI and T-fMRI datasets were substantially different, as anticipated.
• Overlapping ACPs between R-fMRI and T-fMRI datasets indicated subjects not in their expected brain states.
• Analysis of T-fMRI outliers revealed unexpected task performances in some subjects.

Conclusions:

• The novel computational approach effectively differentiates task-free and task-performance brain states using fMRI data.
• The method can identify individuals whose brain activity does not align with their intended task state.
• This work provides new insights into the brain's functional architectures and improves the reliability of fMRI studies.