Metal artifacts reduction (MAR) in dental CT

1. Abstract

In the field of dental and medical radiography, there is an increasing demand for effective metal artifact reduction (MAR) in computed tomography (CT), since the metallic objects-related artifacts seriously degrade the image of CT, resulting in loss of information about teeth and/or other biological structures. We try decomposing the projection data into two parts; the data due to metals only and the background data in the absence of metals. By separating the projections by metals, we can employ sparsity driven metal part reconstruction for metal artifact removal.


2. Metal artifacts

The difference between the X-ray data and Radon transform of the tomography image if represented by









These nonlinearities lead to streaking artifact, which is described in the following figure




















3-1. Method for metal removal and background reconstruction

 

In order to deal with the metal artifacts, we try to decompose the projection data P(x) into two parts:





where w(x) the projection data due to metals only, and u(x) the background projection

in the absence of metals.






























 




1st col.: Original X-ray data (left) and its reconstruction image(right)

2nd col.: Linear interpolation method

3rd col.: Total variation based inpainting method [ref: X. Duan et al, Metal artifact reduction in CT images by sonogram TV inpainting]

4th col.: The proposed method

 

 

3-2. Metal image reconstruction


To overcome the inconsistency between Radon transform of the tomography image and X-ray data, we adopt the following metal part reconstruction problem under the sparsity constraint as in:





















4. Results





































Research Topics

Introduction
Electrical Impedance Tomography
Magnetic Resonance EIT
Electrical Property Imaging
Elastography
LV contours in Ultrasound
Dental CT : Metal Artifacts Reduction
Bioimpedance spectroscopy
Quantitative susceptibility mapping
Image processing
Micro EIT
Surveillance
Lecture Notes
Algorithm
last update 2012.10.26 by Hyungsuk Park