Input functions extraction from gated 18F-FDG PET images

Derivation of the plasma time-activity curve in small animal positron emission tomography (PET) studies is a challenging task. Non-invasive input functions (IF) estimation in cardiac imaging usually involves drawing a region of interest (ROI) within the left ventricle (LV) of the heart. The small size of the LV relative to the resolution of the small-animal PET system, coupled with spillover and heart motion, makes this method difficult. In this work, we acquired rat cardiac images in list-mode with 16 ECG-gates with PET and
¹⁸
F-fluorodeoxyglucose (FDG). We introduced a customized coupled active contour model to reduce the image contamination from blood to tissue and from tissue to blood which are due to organ movements and spillover. The new findings were that we added an external energy to the internal contour to consider the contrast blood-to-tissue as important as the contrast tissue-to-outside myocardium. In order to correct the blood and tissue regions for spillover, we decomposed the two dynamic ROIs in their blood and tissue components using Bayesian probabilities. The results showed a good separation of the blood and tissue components in images as compared to the external blood sampling.