A Reconstruction-Free Projection Selection Procedure for Binary Tomography Using Convolutional Neural Networks

Abstract

In discrete tomography sometimes it is necessary to reduce the number of projections used for reconstructing the image. Earlier, it was shown that the choice of projection angles can significantly influence the quality of the reconstructions. In this study, we apply convolutional neural networks to select projections in order to reconstruct the original images from their sinograms with the smallest possible error. The training of neural networks is generally a time-consuming process, but after the network has been trained, the prediction for a previously unseen input is fast. We trained convolutional neural networks using sinograms as input and the desired, algorithmically determined k-best projections as labels in a supervised setting. We achieved a significantly faster projection selection and only a slight increase in the Relative Mean Error (RME).

Image Quality Improvement in Computed and Binary Tomography

“…The findings of this chapter were in part published in a conference proceedings [84] and in a journal paper [57].…”

“…A negyedik téziscsoportban javasoltam egy eljárást, mely konvolúciós neuronhálók segítségével végez offline vetületválasztást, valamint módosítottam a jól ismert U-Net architektúrát CT képek szegmentálásához. A téziscsoport eredményei egy konferencia kiadványban [84] és egy folyóirat publikációban [57]…”

“…In the fourth thesis group, I proposed a CNN based offline projection selection algorithm and modified the well known U-Net architecture for the segmentation of CT images. The results of this thesis group were published in a conference proceedings [84], and in a journal paper [57].…”