Using normative models pre-trained on cross-sectional data to evaluate intra-individual longitudinal changes in...

Area of Science:

• Neuroimaging
• Brain Development
• Disease Progression

Background:

• Longitudinal neuroimaging is crucial for understanding brain changes over time.
• Current methods often focus on population variation, limiting analysis of individual dynamics.
• A need exists for methodologies that integrate population standards with individual longitudinal changes.

Purpose of the Study:

• To extend the normative modelling framework for analyzing longitudinal neuroimaging data.
• To introduce a quantitative metric (z-diff score) for assessing individual temporal changes against population standards.
• To apply this framework to schizophrenia patients to identify disease-related brain changes.

Main Methods:

• Extended the normative modelling framework to assess longitudinal change relative to population dynamics.
• Developed a 'z-diff' score to quantify individual temporal changes.
• Applied the framework to a longitudinal MRI dataset of 98 early-stage schizophrenia patients.

Main Results:

• The z-diff score revealed a significant normalization of frontal lobe grey matter thickness over one year in schizophrenia patients.
• This normalization was not detected by traditional cross-sectional or longitudinal analyses.
• Cross-sectional analysis showed global grey matter thinning at the initial visit.

Conclusions:

• The proposed framework offers a flexible and effective method for analyzing longitudinal neuroimaging data.
• It provides novel insights into disease progression, particularly for conditions like schizophrenia.
• This approach enhances the understanding of individual brain dynamics in health and disease.