Fluorescence Pharmacokinetic Parametric Imaging Method Based on Dynamic Diffuse Fluorescence Tomography and Deep Learning

At present, the imaging of fluorescence pharmacokinetic parameters based on dynamic diffuse fluorescence tomography (D-DFT) technology have limitations in some aspects including the accuracy of physical model and the quantification of method. In this work, we propose a fluorescence pharmacokinetic parametric imaging method of tumor tissues based on D-DFT and deep learning. It mainly includes: a more realistic training and test simulation data set can be established by combining non-uniform tissue photon transport model and biological tissue fluorescence kinetics method. A fluorescence pharmacokinetic parametric reconstruction algorithm that involves an improved U-Net architecture based on the fully convolutional neural network is developed to break through the complexity bottleneck of imaging physical model. The numerical simulation results show that the method can realize image reconstruction of pharmacokinetic parameters with high spatial resolution and high quantitative accuracy.