Validation of Structural Equation Modeling Methods for Functional MRI Data Acquired in the Human Brainstem and Spinal...

Area of Science:

• Neuroimaging
• Systems Neuroscience
• Biostatistics

Background:

• Functional MRI (fMRI) measures brain activity via blood oxygenation level-dependent (BOLD) signals.
• Structural Equation Modeling (SEM) is a statistical technique for analyzing complex relationships between variables.
• Applying SEM to fMRI data, particularly in the brainstem and spinal cord, requires specialized methods.

Purpose of the Study:

• To adapt and validate SEM for analyzing fMRI data from the brainstem and spinal cord.
• To establish appropriate statistical thresholds and corrections for multiple comparisons in SEM analyses of neuroimaging data.
• To test the efficacy of the adapted SEM methods using existing fMRI datasets of pain processing.

Main Methods:

• Anatomical models were created for thalamus, brainstem, and spinal cord regions involved in pain processing.
• Statistical distributions (Z-scores), significance thresholds, and multiple comparison corrections were determined through simulations using null datasets.
• SEM was applied to fMRI data from healthy participants undergoing noxious stimulation.

Main Results:

• Z-score distributions varied based on the number of regions modeled, time points, and epoch length for dynamic analyses.
• Appropriate statistical thresholds and multiple comparison corrections were successfully demonstrated.
• SEM revealed consistent network features across and within studies, highlighting study condition dependencies.

Conclusions:

• The study successfully adapted and validated SEM for analyzing fMRI data in the brainstem and spinal cord.
• The findings underscore the effectiveness of SEM for uncovering functional neuroanatomy and network dynamics in these critical regions.
• This approach enhances the investigation of pain processing pathways using neuroimaging data.