Modeling venous plasma samples in [¹⁸F] FDG PET studies: a nonlinear mixed-effects approach

Area of Science:

• Nuclear Medicine
• Pharmacokinetics
• Medical Imaging Analysis

Background:

• Dynamic PET imaging typically requires invasive arterial input functions for accurate quantification.
• Image-derived input functions (IDIFs) offer a noninvasive alternative but often need venous plasma samples for rescaling.
• Venous samples are not always obtainable, limiting the application of IDIFs.

Purpose of the Study:

• To develop a nonlinear mixed-effects model to infer venous tracer kinetics.
• To enable reliable dynamic PET quantification using IDIFs when venous samples are missing.
• To integrate population-level venous kinetics with subject-specific covariates.

Main Methods:

• Nonlinear mixed-effects modeling was applied to venous samples from healthy individuals.
• Population parameters (fixed effects) and between-subject variability (random effects) were estimated.
• The influence of subject-specific covariates on tracer kinetics was incorporated.

Main Results:

• A robust model was developed to infer venous tracer kinetics.
• The model successfully estimates kinetics in subjects lacking venous measurements.
• Population parameters and covariate effects were accurately characterized.

Conclusions:

• The developed model facilitates fully noninvasive dynamic FDG PET quantification.
• This method is applicable to both healthy and patient populations with unimpaired hemodynamics.
• It addresses the limitation of missing venous samples in IDIF-based PET analysis.