Kinetic modelling using basis functions derived from two-tissue compartmental models with a plasma input function:...

Area of Science:

• Nuclear medicine
• Pharmacokinetics
• Biophysical modeling

Background:

• Accurate determination of kinetic constants like influx constant (Ki) is crucial for quantitative analysis in positron emission tomography (PET).
• Existing methods for Ki determination have limitations in bias and variability, particularly for tracers like [18F]fluorodeoxyglucose (FDG).

Purpose of the Study:

• To introduce and evaluate a novel kinetic modeling method, basis functions derived from plasma input two-tissue compartmental models (BAFPIC), for determining the influx constant (Ki).
• To compare the performance of BAFPIC against established methods like non-linear least squares (NLLS), autoradiography, and Patlak-Gjedde graphical analysis (PGA).

Main Methods:

• Developed two forms of BAFPIC: BAFPICI (k4=0, no product loss) and BAFPICR (k4 non-zero).
• Simulated kinetic data for [18F]FDG in normal and abnormal brain tissues (homogeneous and heterogeneous).
• Compared BAFPIC with NLLS, autoradiography, and PGA using simulated and real FDG PET data.

Main Results:

• BAFPIC demonstrated superior bias properties compared to NLLS, autoradiography, and PGA for both simulated k4=0 and k4 non-zero data.
• While autoradiography offered the best variability, BAFPICI showed lower variability than PGA and NLLS.
• BAFPICR had inferior variance to PGA but was superior to NLLS for non-zero k4 data.
• Real PET data analysis confirmed BAFPICI's lower noise compared to PGA, with high correlation (r2=0.989) in voxel Ki values.

Conclusions:

• BAFPIC is an effective, easy-to-implement method for voxel-wise Ki determination of [18F]FDG, offering a favorable balance of low bias and good noise characteristics.
• The method is suitable for other tracers following a serial two-tissue compartment model and provides individual kinetic constants and blood volume.
• BAFPIC enhances quantitative accuracy and reduces noise in PET kinetic modeling.