Homomorphic Approach to RF-Inhomogeneity Removal Based on Gabor Filter

Abstract

In this paper a bias correction algorithm for Magnetic Resonance Imaging (MRI) is presented. The Magnetic Resonance (MR) images affected by this artifact, also called RF-Inhomogeneity, exhibit irregular spatial brightness variations caused by magnetic field inhomogeneity. Here we present an original algorithm based on E2D - HUM, already proposed by some of the authors, where a modified Gabor filter is introduced in the elaboration chain to provide directional capabilities to suppress the artifact. The process of restoration doesn't care about the structure of the image and it has been applied to MR images of different parts of body like knee, abdomen, pelvis and brain. A comparison with other algorithms able to remove bias has been made.