PET 18F-FDG kinetic modeling using the iterative two-stage population approach for the assessment of left ventricular function in case of small animal heart failure

Kinetic modeling is widely used to measure micro and macro physiologic parameters from dynamic positron emission tomography (PET). The estimation of the unknown parameters is provided by use of time-activity curves (TACs) calculated from dynamic data and compartmental models. The conventional method used to fit the compartmental models to TACs is the weighted nonlinear least squares (WNLS). This method could be applied either in any given region of interest (ROI) or at the pixel level depending on the radiotracer used. However, in the case of low signal-noise ratio (SNR), WNLS does not yield to a reliable parameter estimation. To overcome this issue, we propose in this paper the iterative two-stage (ITS) method. Unlike WNLS, the ITS method operates all ROIs drawn on PET images of explored organs to estimate parameters in a given ROI. In this work, we study the performance of ITS versus WNLS in the case of heart failure induced by liguature of the left coronary artery. The images are segmented in four tissue regions, that is the anterior wall, the posterior wall, the lateral wall, and the septal wall. In addition, one region was drawn around the blood pool region and used to extract the input function (IF). We conclude that, contrary to the WNLS approach, the ITS approach improves assessment of micro parameters in the septal wall region where SNR is low.