Abstract
231
Objectives Several kinetic models have been applied to hypoxia tracer uptake curves for assessing tumor oxygenation status and have shown advantages in identifying hypoxic regions compared to static analysis. However, the evaluation of the models with respect to the underlying tumor microenvironment is still difficult. We evaluated different kinetic models using digital phantoms of various tumor microenvironments.
Methods 17 digital phantoms (2D, 1x1 mm, pixel: 10x10 μm) of various microvasculature and hypoxia distributions were constructed by simulating the transport and consumption of oxygen, as well as the delivery and the accumulation of 18F-Fmiso tracer using reaction-diffusion equations [1,2]. Kinetic models (Irreversible (Irr.) & reversible (Rev.) two compartment, Thorwarth [3], Patlak, Logan, Cho [4]) were applied to the generated time activity curve (TAC) for each phantom and the correlations (Pearson correlation coefficient) between the kinetic parameters and the underlying vasculature density (VD) and hypoxic fraction (HF) were assessed.
Results Kinetic models can be used to uncover underlying physiologic features. The Cho model is the most sensitive for revealing the hypoxic fraction, followed by the Thorwarth and the Patlak models. The Irr. & Rev. models are the most sensitive for revealing the underlying vasculature density, followed by the Cho model.
Conclusions The Cho model better discriminates the underlying pathophysiological features compared to the other investigated models. Future work will assess the influence of noise and non-specific binding on these models