Quantification of Internal Radioactivity to Measure the Arterial Input Function In Dynamic Positron Emission Tomography: A Simulation Study
Abstract
Purpose
Accurate measurement of the arterial input function (AIF) is important for kinetic modelling in dynamic PET but typically requires invasive arterial blood sampling. This study validates a novel quantification framework that leverages anatomical priors and a single external detector to estimate the AIF non-invasively.
Methods
A 3D radioactivity mapping model was developed wherein gamma camera data is treated as a linear superposition of contributions from n regions of interest (ROIs) defined within a subject-specific “digital twin”. To disentangle these sources, the model analyzes the spatial distribution of detected events using raw spatial moments across m energy windows. These moments are assembled into a global linear system, where the system matrix is derived from Monte Carlo simulations. The system is inverted to solve for the absolute activity of each ROI. Validation was performed using GATE simulations of a realistic digital forearm phantom. A ground-truth AIF was simulated, and the detector performance was optimized by tuning detector and model parameters. Accuracy was assessed by comparing the Area Under the Curve (AUC) of the estimated versus ground-truth AIF.
Results
Detector optimization identified a 25 mm thick LYSO crystal without optical reflectors as the ideal hardware configuration. In parallel, the quantification model performed best when using three energy windows and raw spatial moments up to the third order. With these optimized parameters, the system recovered the AIF with a percent difference in AUC of (1.7 ± 1.8)%.
Conclusion
By incorporating detailed detector physics and subject-specific anatomy within a Monte Carlo–based framework, the proposed method demonstrated accurate AIF estimation using a single external detector. These results support the potential of this approach as a non-invasive alternative to arterial blood sampling and provide a realistic foundation for translation to physical detector systems.