Beyond GLM: Inter-Subject Variability as a Complementary Approach to Detect Longitudinal Changes in Emotion...

Area of Science:

• Neuroimaging
• Neuroscience
• Clinical Psychology

Background:

• Reliably capturing treatment-induced neural changes in neurological conditions is challenging.
• Standard group-level analyses (GLM) in longitudinal neuroimaging studies may obscure individualized neuroplasticity by treating inter-subject variability (ISV) as noise.
• Emotion processing deficits are common in multiple sclerosis (MS) and are a target for rehabilitation.

Purpose of the Study:

• To investigate if modeling inter-subject variability (ISV) can better detect treatment-related neural changes compared to conventional methods.
• To assess the utility of variability-based approaches in capturing neuroplasticity in individuals with neurological conditions undergoing rehabilitation.
• To use emotion-focused rehabilitation as a model to test these neuroimaging analysis techniques.

Main Methods:

• Comparison of General Linear Model (GLM) with threshold-weighted overlap maps (OMth-w) for analyzing neuroimaging data.
• fMRI task involving emotion processing administered to healthy controls (HCs) and people with MS (pwMS) before and after treatment.
• Analysis of spatial consistency across individuals using OMth-w to quantify ISV.

Main Results:

• The standard GLM analysis revealed no significant longitudinal effects in people with MS (pwMS) after treatment.
• The variability-based OMth-w approach detected reduced neural variability in pwMS post-treatment.
• Reduced neural variability correlated with decreased depressive symptoms in pwMS (p < 0.001).

Conclusions:

• Modeling inter-subject variability (ISV) offers a valuable complementary approach to traditional GLM analyses for detecting treatment-related neuroplasticity.
• Variability-based methods can reveal subtle, individualized neural changes missed by group-level analyses in neurological populations.
• This approach enhances the detection of treatment efficacy in conditions like multiple sclerosis, particularly in domains like emotion processing.