Accelerated Parametric Imaging of Dynamic PET using Fully Convolutional Neural Networks

Dynamic positron emission tomography (PET) images are useful in clinical practice because they can be used to calculate the metabolic parameters (Ki) of tissues using graphical methods (such as Patlak plots). Ki is more stable than the standard uptake value (SUV) and has a good reference value for clinical diagnosis. However, the long scanning time required for obtaining dynamic PET images, usually an hour, makes this method less useful in some ways. There is a tradeoff between the scan durations and the signal-to-noise ratios (SNRs) of Ki images. The purpose of our study is to obtain approximately the same image as that produced by scanning for one hour in just half an hour, improving the SNRs of images obtained by scanning for 30 minutes and reducing the necessary 1-hour scanning time for acquiring dynamic PET images. In this paper, we present a new approach to parametric imaging in positron emission tomography (PET) using a fully convolutional neural network with a U-Net architecture. Our method significantly reduces scanning time, allowing for higher-quality parametric images to be obtained after scanning for only 30 minutes compared to the standard hour-long scanning time. We also introduce a novel kinetic model network for processing dynamic PET images. Our proposed approach improves the signal-to-noise ratio and is more robust than previous methods that use the Patlak plot for quantitative analysis. Our method also reduces the noise and generation time of parametric images, making them available for clinical use more quickly. We demonstrate the efficacy of our method with simulations and real patient data.