Affiliations: Department of Radiation Oncology, Stanford University
School of Medicine, Stanford, CA 94305-5847, USA | Department of Radiation Oncology, UT Southwestern
Medical Center, Dallas, TX 75390-9183, USA
Abstract: The presence of metals in patients causes streaking artifacts in
X-ray CT and has been recognized as a problem that limits various applications
of CT imaging. Accurate localization of metals in CT images is a critical step
for metal artifacts reduction in CT imaging and many practical applications of
CT images. The purpose of this work is to develop a method of
auto-determination of the shape and location of metallic object(s) in the image
space. The proposed method is based on the fact that when a metal object is
present in a patient, a CT image can be divided into two prominent components:
high density metal and low density normal tissues. This prior knowledge is
incorporated into an objective function as the regularization term whose role
is to encourage the solution to take a form of two intensity levels. A computer
simulation study and four experimental studies are performed to evaluate the
proposed approach. Both simulation and experimental studies show that the
presented algorithm works well even in the presence of complicated shaped metal
objects. For a hexagonally shaped metal embedded in a water phantom, for
example, it is found that the accuracy of metal reconstruction is within sub-millimeter.
Keywords: Computed tomography, metal artifacts, gradient-controlled penalty, and iterative image reconstruction
