An input function estimation method for dynamic mouse 18F-FDG microPET studies

We present and validate a method to estimate the input function (IF) from dynamic mouse ¹⁸F-FDG microPET images and 1 late blood sample. The proposed method is almost entirely noninvasive, and accounts for the spillover, partial-volume, delay and dispersion effects. First, the time-activity curves (TACs) of the left ventricle (LV), the myocardium and the liver were extracted from their respective feature extractuion(VOIs). The liver TAC data points between 35 seconds and 1500 seconds were used to substitute for blood samples since we found they were highly correlated (average correlation coefficient r = 0.989 ± 0.012). The IF to be estimated (EIF) was expressed as a mathematical model in which the parameters were simultaneously estimated by fitting the modeled LV and myocardium TACs to the mouse PET data for these organs. Twenty normal mice data sets from the Mouse Quantitation Program database, which were shared by UCLA, were used to verify our method. The differences of the area under the curves between the EIF and the true IF (blood samples) was 6.9% ± 13.2%, and the difference of the 18F-FDG influx constant K_i in myocardium was 4.8% ± 16.7% (r = 0.931). The experimental results demonstrate the effectiveness of the proposed method.