Initial Image Selection In Limited Angle Tomographic Imaging

Abstract

In limited angle tomographic imaging, artifacts arise due to missing data during the acquisition. To deal with this problem, iterative image reconstruction algorithms have been developed. In iterative reconstruction algorithms, the initial image guess which is often neglected is very crucial and plays an important role as it directly affects the convergence rate. This paper presents a comparison of three different initial images (zeroes image, average image, filtered back projected image) to observe their contribution to the convergence rate. Computer simulations are performed by using algebraic reconstruction technique with total variation, (ART + TV). Root means squared error (RMSE) of a specific layer of interest (LOI) of a 3D phantom is used to compare the impact of appropriate prior image selection. For a better visual observation, structure similarity (SSIM) parameters are also compared.