Task activations produce spurious but systematic inflation of task functional connectivity estimates

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Neuroimaging

Background:

• Neuroscientific studies often focus on task-evoked activations, limiting understanding of brain region interactions.
• Task-state functional connectivity (FC) quantifies these interactions but is often confounded by task activations.
• Current methods may lead to ambiguous interpretations of brain activity during tasks.

Purpose of the Study:

• To investigate the impact of task-evoked activations on task-state FC estimation.
• To evaluate methods for correcting activation-induced inflation in FC.
• To determine optimal preprocessing steps for accurate task-state FC analysis.

Main Methods:

• Utilized a neural mass computational model to simulate fMRI data.
• Assessed the influence of task-evoked activations on FC estimates (Pearson correlation, PPI).
• Compared standard activation removal techniques with methods that flexibly model response shapes.

Main Results:

• Task-evoked activations significantly inflate task-state FC estimates, particularly in fMRI data.
• Standard methods for removing activations were largely ineffective at correcting this inflation.
• Flexible modeling of task-evoked response shapes successfully corrected FC inflation without removing true effects.

Conclusions:

• Task-evoked activations pose a significant challenge for accurate task-state FC estimation.
• Flexible modeling of task response shapes is a crucial preprocessing step for valid FC analysis.
• This approach enhances the reliability of neuroimaging findings on brain connectivity during tasks.