Comparison of algebraic reconstruction techniques for tomosynthesis

Digital Breast Tomosynthesis (DBT) is an innovative 3D imaging technique implemented using a limited number of low dose projections, which are taken with the x-ray source moving in a limited angle of rotation around the breast. These low dose projection images should become processed using mathematical methods, to reconstruct tomographic images, resulting in a 3D representation of the imaged breast. In tomosynthesis imaging, out-of-focus slice blur problem arises due to incomplete sampling problem. Several approaches have been suggested to deal with this problem. The purpose of this work is to implement Algebraic Reconstruction Technique (ART) and Multiplicative Algebraic Reconstruction Technique (MART) which are kind of that approach. Former studies addressed largely 2D tomosynthesis image reconstruction problem. In this study, a 3D phantom model was used. All of the algorithms and experiments are programmed with C++. The differences between two reconstruction algorithms are also investigated by means of comparing root mean square error (RMSE) and mean structural similarity (MSSIM).