When most fMRI connectivity cannot be detected: Insights from time course reliability.

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Brain Connectivity Analysis

Background:

• Accurate measurement is crucial for determining correlations between phenomena.
• Functional MRI (fMRI) connectivity studies often face challenges in understanding the impact of within-subject reliability on detectable signal strength.
• The precise limitation imposed by within-subject time course reliability on group-level connectivity detection remains unclear.

Purpose of the Study:

• To investigate how within-subject time course reliability affects the detection of fMRI connectivity.
• To quantify the proportion of detectable connectivity explained by cognitive tasks, specifically working memory.
• To assess the influence of within-subject reliability on the statistical significance of group-level connectivity findings.

Main Methods:

• Estimated functional connectivity from fMRI data acquired during a working memory task.
• Calculated grand mean connectivity for test and retest runs.
• Incorporated test-retest reliability and auto-correlations of time courses to determine detectable connectivity.
• Utilized Null Hypothesis Significance Testing (NHST) to evaluate group-level detection significance.

Main Results:

• Grand mean connectivity was high (r ≈ 0.40) in both test and retest runs.
• When within-subject reliability was considered, mean detectable connectivity dropped significantly to r = 0.09, suggesting less than a quarter of the initial connectivity is reliably detectable.
• Only 0.81% of the detectable connectivity was explained by working memory task-related time course fluctuations.
• Within-subject reliability critically impacted statistical significance at the group level, especially with small sample sizes or random coordinate selection.
• A sample of 50 individuals showed perfect reproduction of the connectome between sessions (P < 2.54e-6) despite minimal cognitive explanation.

Conclusions:

• Within-subject time course reliability is a major constraint on the detection of true functional connectivity in fMRI.
• Observed connectivity in fMRI studies may substantially overestimate the influence of cognitive processes due to unreliability.
• Assessing within-subject time course reliability is essential for accurate interpretation of fMRI connectivity findings and should be conducted more routinely.