Chinese Journal of Computers   Full Text
Title	A Novel Algorithm to Reduce the Gibbs Ringing Artifacts in Vivo MRI
Authors	JIANG Gui-Ping HUANG Xin FENG Yan-Qiu CHEN Wu-Fan
Address	(Institute of Medical Information Processing, School of Biomedical Engineering, Southern Medical University, Guangzhou 510515)
Year	2007
Issue	No.11(2040—2047)
Abstract &   Background	Abstract Gibbs ringing in magnetic resonance imaging is a well know artifact which is prevalent particularly at the tissue boundaries, this phenomena results from the reconstruction procedure involving only part of the k-space data. The Gegenbauer reconstruction method has been shown to be able to eliminate Gibbs artifacts effectively while retaining high resolution. Its disadvantages include time-consuming and the reconstruction result depending on the selection of parameters greatly. In this paper, the authors improve the Gegenbauer method by introducing the Inverse Polynomial Reconstruction Method (IPRM) and replacing the Gegenbauer polynomial with Chebyshev polynomial. The new method reduces the construction error and computational cost effectively without any need to select the parameters. Because the method above is discussed in smooth interval, the edge detection becomes critical in determining the smooth intervals for high resolution reconstruction. This paper presents an edge detection method which can achieve precise edge effectively and make the new reconstruction method suitable both in vitro and vivo. keywords magnetic resonance imaging; Gibbs ring artifact; Chebyshev polynomial; inverse polynomial reconstruction; edge detection background This research is supported by the National Basic Research Program(973 Program) of China under grant No.2003CB716100—Research of Key Scientific Problems in Clinical Medical Information Processing called MIP973. The MIP973 aims to establish a set of fundamental theory and algorithms for medical image processing and electrophysiological signal processing, which are one of the most essential technical supports for the development of medical imaging equipments and electrophysiological devices. It will organize an outstanding scientific team to solve the key problems as follows: Stochastic modeling from some concrete problems to general problems, unique property of the solution under the conditions of optimal constraint, iterative stability and rapid convergence for adaptively estimating various parameters. At the same time, the project will establish multimodality information registration model, area segmentation model, artifact correct model, optimal constraint model, optimal searching strategy, and so on. The promise of this project is not only enriches the content of life science and information science to promote the development of these fields, but also becomes a kind of new knowledge economy in post-processing software of medical clinical information. The MIP973 is divided into 6 subprojects. This paper is an outcome of the image reconstruction group, one direction of the first subproject—Stochastic Models and Optimal Algorithms of Medical Information Processing. The objective of the group is to study how to improve the performance of image reconstruction and artifact correction. In the past years, the group has done a lot of work on this area, for example, presenting a new algorithm for extracting motion information from PROPELLER data and head motion correction in T1-Weighted MRI. For the reconstruction of PROPLLER data, algorithms to reliably and accurately extract inter-strip motion from data in central overlapped area are crucial to motion artifacts suppression. When implemented on T1-weighted MR data, the reconstruction algorithm, with motion estimated by registration based on maximizing correlation energy in frequency domain(CF), produces images with low quality due to the inaccurate estimation of motion. In this paper, a new algorithm is proposed for motion estimation based on the registration by maximizing mutual information in spatial domain(MIS). Furthermore, the optimization process is initialized by CF algorithm, so the algorithm is abbreviated as CF-MIS algorithm in this paper. With phantom and in vivo MR imaging, the CF-MIS algorithm is shown to be of higher accuracy in rotation estimation than CF algorithm. Consequently, the head motion in T1-weighted PROPELLER MRI is better corrected.