Summary:

Experimental Design:

1. ROI Analysis Group:
• 14 healthy subjects (ages 18-25, mean 21.2 years, 7 females)
• Performed three functional tasks followed by 4-minute task-free scan (eyes closed)
• No behavioral assessments collected for this group
2. ICA Analysis Group:
• 21 healthy subjects (ages 18-70, mean 34.6 years, 11 females)
• 5-minute task-free scan (eyes closed)
• 15 subjects completed prescan anxiety ratings (Visual Analog Scale 0-10)
• All 21 completed neuropsychological testing (Trail Making Test)

Analysis Methods:

1. ROI-Based Functional Connectivity:
• Two seed ROIs selected from previous two-back working memory task: right DLPFC (BA 45/46) and right FI (BA 47/12)
• Extracted average time series from each ROI
• Preprocessing: global drift scaling and bandpass filtering (0.0083-0.15 Hz)
• Linear regression analysis with white matter nuisance covariates
• Random-effects group analysis (p < 0.001 height and extent, corrected)
2. Independent Component Analysis:
• Used FSL MELODIC software
• Automated three-step component selection:
• Frequency filter to remove high-frequency components (>0.1 Hz = 50%+ power)
• Template-matching using ROI-derived networks as templates
• Selected best-fit component for each subject
• No bandpass filtering (ICA isolates noise as independent components)
3. Behavioral Correlation Analyses:
• Tested correlations between network connectivity and:
• Prescan anxiety (VAS ratings)
• Executive function (Trail Making Test B - A time)
• Covariate-of-interest analyses masked to respective networks
• Statistical threshold: p < 0.01 height and extent, corrected