The λ-MLEM Algorithm: An Iterative Reconstruction Technique for Metal Artifact Reduction in CT Images

Abstract

Filtered backprojection (FBP) is an inadequate method to cope with inconsistencies in Radon space and, consequently, leads to artifacts in reconstructed CT images. A solution to this problem is given by statistical reconstruction methods like the Maximum-Likelihood Expectation-Maximization (MLEM) algorithm. The advantage of MLEM is that it allows to weight raw projection data during reconstruction. The method presented here consists of two steps. In a first step, inconsistent data in the Radon space were bridged using a directional interpolation scheme. Since these surrogate data are contaminated with residual inconsistencies, in a second step, the image is reconstructed using a weighted MLEM algorithm. In this work, the modified MLEM algorithm for metal artifact reduction in CT is presented for clinical hip prosthesis data. On the basis of image entropy the reconstruction success is evaluated.