A new framework for metabolic connectivity mapping using bolus [¹⁸F]FDG PET and kinetic modeling

Area of Science:

• Neuroimaging
• Metabolic Imaging
• Positron Emission Tomography (PET)

Background:

• Metabolic connectivity (MC) is often studied as across-individual (ai-MC) covariation in static [18F]FDG PET images.
• Within-individual MC (wi-MC) from dynamic [18F]FDG PET signals is less explored, raising questions about validity and interpretability.
• Comparing ai-MC and wi-MC with functional connectivity (FC) and structural connectivity is crucial for understanding brain networks.

Purpose of the Study:

• To develop a novel methodology for calculating within-individual metabolic connectivity (wi-MC) using dynamic [18F]FDG PET data.
• To compare ai-MC derived from Standardized Uptake Value Ratio (SUVR) and kinetic parameters (Ki, K1, k3) of [18F]FDG.
• To assess the interpretability of MC by comparing wi-MC and ai-MC with structural connectivity and resting-state fMRI FC.

Main Methods:

• Developed a novel wi-MC approach utilizing Euclidean distance on PET time-activity curves.
• Calculated ai-MC using [18F]FDG kinetic parameters (Ki, K1, k3) and SUVR.
• Compared wi-MC and ai-MC matrices with each other and with fMRI FC and structural connectivity data.

Main Results:

• Across-individual MC maps differed based on the [18F]FDG parameter used (e.g., k3 MC vs. SUVR MC, r=0.44).
• wi-MC and ai-MC matrices showed limited similarity (maximum r=0.37).
• wi-MC demonstrated higher similarity to fMRI FC (Dice similarity: 0.47-0.63) compared to ai-MC (0.24-0.39).

Conclusions:

• Calculating individual-level metabolic connectivity from dynamic PET is feasible using the novel Euclidean distance method.
• wi-MC matrices derived from dynamic PET are interpretable and show greater similarity to functional connectivity than ai-MC.
• This study validates wi-MC as a valuable approach for investigating brain network dynamics using PET imaging.