Main Article Content
This work is licensed under a Creative Commons Attribution 4.0 International License.
The acquisition views of projection magnetic particle imaging (MPI) limit the temporal resolution of tomographic imaging. When the views of projection are insufficient (sparse view), the streaking artifacts will be introduced after filtered back-projection. The current solutions to the sparse view problem in computed tomography can be divided into three categories, iterative reconstruction, image post-processing, and sparse view sinogram restoration. The first one is computationally intensive and the parameters are difficult to determine. The latter two are data-driven deep learning methods that require large-scale trainable datasets. However, the complex features of the image domain will make the network easy to overfit. Therefore, we propose a sparse view sinogram restoration network for MPI to improve the temporal resolution of tomography. We validate the effectiveness of the proposed method on a simulated dataset and outperform iterative reconstruction and image post-processing methods.