Population-based model selection for an accurate estimation of time-integrated activity using non-linear...

Area of Science:

• Nuclear Medicine
• Radiotherapy Physics
• Pharmacokinetics

Background:

• Accurate absorbed dose determination is crucial for personalized treatment planning in Molecular Radiotherapy (MRT).
• Calculating the absorbed dose relies on Time-Integrated Activity (TIA) and dose conversion factors.
• Selecting the appropriate fit function for TIA calculation remains a challenge in MRT dosimetry.

Purpose of the Study:

• To develop and evaluate a data-driven method for accurate TIA determination in MRT.
• To implement a Population-Based Model Selection within the Non-Linear Mixed-Effects (NLME-PBMS) framework.
• To address the unresolved issue of fit function selection for TIA calculations in MRT.

Main Methods:

• Utilized biokinetic data of a Prostate-Specific Membrane Antigen (PSMA) radioligand for cancer treatment.
• Derived and fitted eleven mono-, bi-, and tri-exponential functions using the Non-Linear Mixed-Effects (NLME) framework.
• Employed Akaike weights for fit function selection and performed NLME-PBMS Model Averaging (MA) for robust TIA calculation.

Main Results:

• The function [Formula: see text] was identified as the most supported fit function with an Akaike weight of (54 ± 11)%.
• The NLME model selection method demonstrated superior or equivalent performance compared to Individual-Based Model Selection (IBMS) and Shared-Parameter Population-Based Model Selection (SP-PBMS).
• Achieved significantly lower Root-Mean-Square Errors (RMSE) for TIA calculations: 2.4% for NLME-PBMS (f_3a) versus 7.4% for IBMS and 8.8% for SP-PBMS.

Conclusions:

• A novel procedure was developed for selecting the optimal fit function for TIA calculation in MRT using population-based methods.
• This technique integrates Akaike-weight-based model selection with the NLME framework, standard practices in pharmacokinetics.
• The developed method enhances the accuracy of TIA determination, supporting more precise absorbed dose calculations for MRT.