Uncertainties in compartmental kinetic analysis of moving targets

Objectives This study estimates the breathing-induced uncertainties in the kinetic parameters of an ¹⁸F-FMISO hypoxia two-tissue compartment model.

Methods Spheres with diameters ranging between 2 cm and 6 cm were simulated using the GEANT4 Application for tomographic emission (GATE) package. Each of the spheres was simulated in both static and oscillatory modes with a 5 sec period and motion amplitudes of 1 cm, 2cm, and 4 cm, to simulate breathing. The activity concentrations of the target and the background at each time frame of dynamic PET images were simulated following the time activity curve (TAC) extracted from one clinical NSCLC FMISO dynamic data set. Kinetic analysis for each of the simulated dynamic PET data sets was conducted using PMOD software, and the uncertainties in the kinetic rate constants due to target motion were assessed by comparison with those deduced from the corresponding static data sets. Specific attention is paid to the perfusion rate, k1, and trapping rate, k3, which correlates with hypoxia.

Results For target sizes in the order of 2 cm in diameter, a reduction of ~15% and ~31% errors in k3, and an increase of ~12% and ~23% errors in k1 were observed for motion amplitudes of 0.5 cm and 1.5 cm respectively. For larger target sizes, e.g. 4 cm in diameter, those errors are reduced to ~6% and ~11% in k3, and ~4% and 5% in k1 were observed for motion amplitudes of 0.5 cm and 1.5 cm respectively.

Conclusions Target motion can be a major source of error in compartmental kinetic analysis of moving target. It yields increase in the apparent tumor perfusion and a reduction in the apparent trapping rate. Those pitfalls become particularly pronounced when the motion amplitude is comparable to target size. Motion correction in the dynamic PET images may therefore become a pre-requisite for compartmental kinetic analysis of moving targets.