Diagnostically distinct resting state fMRI energy distributions: A subject-specific maximum entropy modeling study

Area of Science:

• Neuroimaging
• Psychiatry
• Computational Neuroscience

Background:

• Traditional neuroimaging studies analyze brain activation (first-order) and functional connectivity (second-order).
• These methods have limitations; integrating them via maximum entropy modeling (MEM) offers a more comprehensive view of fMRI data.
• MEM provides a system-wide measure called 'energy' for brain activity patterns.

Purpose of the Study:

• To assess the applicability of individual-level MEM for psychiatric disorders.
• To identify image-derived model coefficients linked to MEM parameters in clinical populations.
• To explore how MEM-derived brain energy measures differentiate psychiatric conditions.

Main Methods:

• Resting-state fMRI data from UK Biobank participants (n=264) with schizophrenia/schizoaffective disorder, bipolar disorder, major depression, and healthy controls were analyzed.
• Pairwise maximum entropy models (MEMs) were fitted to subsets of the default mode network (DMN).
• Model parameters were correlated with image-derived coefficients.

Main Results:

• The MEM successfully explained brain energy state probabilities across all participants.
• Model parameters significantly correlated with image-derived coefficients in all groups.
• Averaged energy levels were higher in schizophrenia and major depression, but lower in bipolar disorder compared to controls within the DMN.

Conclusions:

• Distinct energy states identified by MEM suggest unique temporal dynamics underlying different psychiatric diagnoses.
• Subject-specific MEMs account for individual variations, providing improved biologically meaningful correlates of brain activity.
• This approach holds potential clinical utility for understanding and potentially diagnosing psychiatric disorders.