Abstract: With the vigorous development and wide application of positron emission tomography (PET), its reconstruction algorithms have been significantly improved. Recently, deep learning (DL) has demonstrated excellent learning ability in high-dimensional and highly complex data, and shows great potential in improving the speed and accuracy of PET image reconstruction.Several papers have confirmed its promising development in PET image reconstruction. This paper reviews the latest research progress of DL in static PET image reconstruction at home and abroad.