Bayesian Estimation of the ntPET Model in Single-Scan Competition PET Studies

Area of Science:

• Nuclear Medicine
• Pharmacology
• Biophysics

Background:

• Positron Emission Tomography (PET) is crucial for studying neurotransmitter systems and drug interactions in vivo.
• Existing PET kinetic modeling methods, like lp-ntPET, face limitations in accuracy and noise sensitivity due to discretized basis functions.
• Accurate quantification of endogenous neurotransmitter release and drug displacement is essential for understanding neurological disorders and drug efficacy.

Purpose of the Study:

• To develop and validate an innovative Bayesian-based nonlinear PET (b-ntPET) method for improved analysis of competition models.
• To enhance the sensitivity, accuracy, and robustness of PET kinetic modeling, particularly in scenarios involving endogenous neurotransmitter release or drug competition.
• To enable more efficient detection and characterization of dynamic drug occupancy from single PET scans.

Main Methods:

• Developed b-ntPET, a novel resolution approach utilizing Markov-Chain Monte-Carlo (MCMC) sampling to estimate parameter probability distributions.
• Integrated joint estimation of model parameters and noise levels for increased robustness.
• Employed a Bayesian framework to incorporate prior knowledge, guiding the estimation process toward realistic solutions.

Main Results:

• Simulations demonstrated that b-ntPET is substantially more robust to noise, sensitive, and accurate compared to the reference lp-ntPET method.
• In animal studies with [18F]MPPF, b-ntPET successfully measured dose-dependent responses to a competing agonist, a feat where lp-ntPET failed.
• In human [11C]raclopride studies, b-ntPET detected significant endogenous dopamine releases induced by transcranial direct current stimulation, unlike lp-ntPET.

Conclusions:

• The b-ntPET method demonstrates superior noise resistance and robustness across various experimental contexts compared to lp-ntPET.
• b-ntPET effectively detects and characterizes dynamic drug occupancy, offering improved performance in pharmacological challenge studies and endogenous release detection.
• This advancement facilitates more efficient and reliable analysis of PET data for drug development and neuroscience research.