Generation of Whole-Body FDG Parametric Ki Images From Static PET Images Using Deep Learning

F-fluorodeoxyglucose parametric
$K_{\mathrm{ i}}$
images show a great advantage over static standard uptake value (SUV) images, due to the higher contrast and better accuracy in tracer uptake rate estimation. In this study, we explored the feasibility of generating synthetic
$K_{\mathrm{ i}}$
images from static SUV ratio (SUVR) images using three configurations of U-Nets with different sets of input and output image patches, which were the U-Nets with single input and single output (SISO), multiple inputs and single output (MISO), and single input and multiple outputs (SIMO). SUVR images were generated by averaging three 5-min dynamic SUV frames starting at 60-min post-injection, and then normalized by the mean SUV values in the blood pool. The corresponding ground-truth
$K_{\mathrm{ i}}$
images were derived using Patlak graphical analysis with input functions from the measurement of arterial blood samples. Even though the synthetic
$K_{\mathrm{ i}}$
values were not quantitatively accurate compared with ground truth, the linear regression analysis of joint histograms in the voxels of body regions showed that the mean
$R^{2}$
values were higher between U-Net prediction and ground truth (0.596, 0.580, and 0.576 in SISO, MISO, and SIMO), than that between SUVR and ground truth
$K_{\mathrm{ i}}$
(0.571). In terms of similarity metrics, the synthetic
$K_{\mathrm{ i}}$
images were closer to the ground-truth
$K_{\mathrm{ i}}$
images (mean SSIM = 0.729, 0.704, and 0.704 in SISO, MISO, and MISO) than the input SUVR images (mean SSIM = 0.691). Therefore, it is feasible to use deep learning networks to estimate the surrogate map of parametric
$K_{\mathrm{ i}}$
images from static SUVR images.