Calculation of Projection Matrix in Image Reconstruction Based on Neural Network

Abstract

The iterative reconstruction algorithms can get better reconstruction result by adding constraints in the case of incomplete or uneven projection. In the iterative reconstruction algorithm, how to obtain the relationship between the reconstructed image and the projected data, namely, projection matrix, is the key to the image reconstruction. In this paper, the neural network algorithm is used to calculate the projection matrix, which gives a solution to a class of problems. In simulation, we use the projection matrix trained by the neural network to achieve the reconstruction, and the results show that the original image can be well reconstructed.