Deep unrolled primal dual network for TOF-PET list-mode image reconstruction

<i>Objective.</i>Time-of-flight (TOF) information provides more accurate location data for annihilation photons, thereby enhancing the quality of positron emission tomography (PET) reconstruction images and reducing noise. List-mode reconstruction has a significant advantage in handling TOF information. However, current advanced TOF-PET list-mode reconstruction algorithms still require improvements when dealing with low-count data. Deep learning algorithms have shown promising results in PET image reconstruction. Nevertheless, the incorporation of TOF information poses significant challenges related to the storage space required by deep learning methods, particularly for the advanced deep unrolled methods.<i>Approach.</i>In this study, we propose LMPDnet, a deep unrolled primal dual network for TOF-PET list-mode reconstruction. The network is unrolled into multiple phases, with each phase comprising a dual network for list-mode domain updates and a primal network for image domain updates. We utilize CUDA for parallel acceleration and computation of the system matrix for TOF list-mode data.<i>Main results.</i>Reconstructed images of different TOF resolutions and different count levels show that the proposed method shows better noise suppression and image quality compared to the list-mode ordered subset expectation maximization, total-variation regularized list-mode expectation maximization, list-mode stochastic primal dual hybrid gradient, total-variation regularized stochastic primal dual hybrid gradient and HistoCNN-2D (similar to FastPET by replacing 3D convolutions with 2D) method in both visually and quantitative analysis.<i>Significance.</i>The results demonstrate the potential application of deep unrolled methods for TOF-PET list-mode data and show better performance than current mainstream TOF-PET list-mode reconstruction algorithms, providing new insights for the application of deep learning methods in TOF list-mode data. The codes for this work are available athttps://github.com/RickHH/LMPDnet.