A New Algorithm Of Gibbs Artifacts Removing In MR Images

With the development of medical level and computer technology, magnetic resonance imaging technique is becoming one indispensable part of disease diagnosis. Unfortunately, MR imaging time is too long to satisfy the need for some special patients, such as bad or incoordinate patients. In practice, the number of phase-encoded signals is often reduced to minimize the acquisition time; thereby we only collected partial raw data. However, reduction of the number of phase-encoded signals results in the well-known Gibbs oscillations when the truncated k-space data is Fourier transformed to reconstruct the image.The Gibbs phenomenon has long been a hindrance to the image reconstruction. So far some methods to reduce Gibbs phenomenon have been proposed. Filters are often introduced to reduce the effects of Gibbs ringing, but filtering inevitably causes blurring of the features of the reconstructed image at the boundaries. In the recent years, there are urgent needs as to how to remove Gibbs ringing artifact in a magnetic resonance imaging research. The purpose of this study is to reduce the Gibbs ringing artifact and improve the quality of MR imaging with partial raw data. In this paper, a novel and accurate algorithm, namely the inverse polynomial reconstruction method (IPRM) based on Chebyshev, is proposed to eliminate Gibbs artifact.Gegenbauer reconstruction method was proposed in recent years, and has been shown to effectively eliminate the effects of Gibbs phenomenon. The remarkable advantage of this method is to eliminate the Gibbs ringing artifact without compromising high resolution at the edges, which it is difficult to solve for common artifact removing technique. In this paper, we analyze theory of this algorithm and discuss the limitation. From the experiments, we draw the conclusion that the key problem on the Gegenbauer reconstruction method is that the condition on parametersis sought in order to obtain spectral convergence, and if not adequate, heavily influence the quality of reconstructed images.To solve these problems, we introduce the inverse polynomial reconstruction method (IPRM). It improves the Gegenbauer reconstruction method and solves the parameters problems that influence on the quality of reconstructed images, which in turn IPRM decrease the reconstructed error. At the same time, we replace the Gegenbauer polynomial with Chebyshev polynomial in this paper, which exempt from choose of one of the parameters. We effectively achieve shorter reconstructed time without compromising high resolution.In addition, because the above method is discussed in smooth interval, we must detect edge information with the spatial frequency domain (k-space) data and obtain a series of smooth intervals. Clearly, the edge detection becomes critical in determining the smooth intervals for high resolution reconstruction. In the clinical application, complicated body structure give rise to intricate edge information of reconstructed images. This paper presents an edge detection method with frequency filtering, which is shown effectively eliminate the influence of Gibbs ringing. Fortunately, moreover, such a procedure is computationally efficient.This paper presents an edge detection method which can effectively achieve precise edge and make IP based on Chebyshev reconstruction method suitable for complicated real partial MR data of human body. The proposed method is verified with experiment of artifact removal. This paper demonstrates the advantages of using the proposed reconstructed algorithm by comparing it to the common method on phantom and medical MR images.