Research On Improved Methods For CT Ring Artifact Reduction

CT(Computed Tomography) technology can not only show the internal details of the object structure clearly, but also give details of material quantitatively. It has been used in industrial imperfection detection, medical diagnosis, and the measurement of properties. Ring artifacts are caused by bad pixel channel. It makes CT image difficult to identify and impedes post processing. It could lead to wrongly diagnosis because pathology image has similar characteristics with ring artifact. Therefore, removing such artifact is indispensable.According to ring artifact characteristics in projection sinogram image and reconstruction image by studying the causes, two kinds of ring artifact correction methods are proposed in the paper: ring artifact correction based on projection sinogram with subset differential and ring artifact correction based on polar coordinates with improved two dimension low pass filter.Ring artifact correction based on projection sinogram with subset differential is a pre-processing method. It uses spatial differential after sum up pixels of all lines separately when extracts artifact information, it avoids choosing threshold value with traditional method. Since artifact datum is processed separately, it can avoid Adjacent data interference in interpolation method. The most important thing is, This method can process the preliminary processed image by separating the data to subsets to remove the remain arc artifact data accurately.Ring artifact correction based on polar coordinates with improved two dimension low pass filter is a post-processing method. It can protect the detail information in original image when filtering the artifact information with common two dimension low pass filter by designing a two dimension low pass filter with a slit. Through coordinates transform, using t he designed filter to process the artifact image in frequency domain. It can protect the original detail information and remove the artifact information as much as possible. After Fourier inverse transformation and coordinate inverse transformation, the artifacts can be removed.Using the above two correction methods for laboratory simulation and real CT images in experiments, from the results, we can clearly see that two kinds of methods can effectively remove the ring artifacts of CT images. The method based on projection sinogram processes the artifact data before reconstruction, so it can promise a better resolution of processed CT image. The method based on polar coordinates can remove the artifact information well when the projection data can not be obtained.