Alzheimer's Imaging Consortium.

Area of Science:

• Neuroimaging
• Network Neuroscience
• Alzheimer's Disease Research

Background:

• Positron Emission Tomography (PET) neuroimaging quantifies in vivo amyloid and tau accumulation.
• Network approaches are valuable for studying brain structure and function in Alzheimer's Disease (AD).
• Estimating participant-level networks from static PET data has been challenging, hindering multimodal integration.

Purpose of the Study:

• To develop a novel framework for deriving participant-level tau similarity networks from PI-2620 tau PET data.
• To analyze tau accumulation patterns in a large, diverse cohort (HABS-HD study).
• To explore the utility of these networks for understanding AD neurobiology and enabling multimodal integration.

Main Methods:

• Extracted Standardized Uptake Value Ratios (SUVr) from 100 cortical regions (Schaefer parcellation).
• Generated participant-level similarity networks (1-abs(SUVr(i)-SUVr(j))) and reciprocal of absolute difference (RAD) networks (1/abs(SUVr(i)-SUVr(j))).
• Compared these networks to intersubject correlation-based covariance networks and stratified participants into Tau+ and Tau- subgroups.

Main Results:

• All three network types exhibited block structures within canonical resting state networks.
• Moderate correlations were found between covariance networks and the new participant-level network types.
• Tau positivity stratification revealed nearly identical similarity networks but moderately correlated RAD networks between Tau+ and Tau- groups.

Conclusions:

• Participant-level tau PET networks, particularly similarity networks, can be normalized for topological analysis, enhancing AD neurobiology understanding.
• RAD networks preserve SUVr differences, facilitating patient-centric multimodal network approaches.
• These PET networks show potential for integration with MRI data for precision diagnostics and AD subtyping.