Hindawi Publishing Corporation EURASIP Journal on Advances in Signal Processing Volume 2011, Article ID 183547, 13 pages doi:10.1155/2011/183547 Research Article A Self-Adaptive Approach for the Detection and Correction of Stripes in the Sinogram: Suppression of Ring Artifacts in CT Imaging A. N. M. Ashrafuzzaman, 1 Soo Yeol Lee, 2 and Md. Kamrul Hasan 1, 2 1 Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh 2 Department of Biomedical Engineering, Kyung Hee University, Kyungki 446-701, Republic of Korea Correspondence should be addressed to Md. Kamrul Hasan, khasan@eee.buet.ac.bd Received 4 June 2010; Revised 2 August 2010; Accepted 14 August 2010 Academic Editor: Antonio Napolitano Copyright © 2011 A. N. M. Ashrafuzzaman et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The digital X-ray detectors often generate stripe artifact in the sinogram which in turn creates ring artifact in the reconstructed micro-Computed Tomography (μ-CT), C-Arm CT, and most recent dental CT images. Such ring artifacts not only obscure image details in the regions of interest but also mask the whole image with some artifacts. In this paper, novel techniques are proposed for the detection and suppression of ring artifacts in the sinogram domain. As ring artifacts are manifested as edge creating stripes, single or contiguous, in the sinogram, they are detected based on a set of specific conditions derived from the second derivative of the sinogram and a new self-adaptive threshold computed from its first derivative. A new method for the detection of wide band contiguous stripes using the mean curve and multilevel polyphase decomposition of the given sinogram is also proposed here. For the correction of ring artifacts, novel variable window moving average (VWMA) and weighted moving average (WMA) filters are proposed in this work. To evaluate and compare the performance of the proposed algorithm, various types of synthetic and real μ-CT images are used. Experimental results show that the proposed method can detect ring artifacts with high accuracy and thus remove them more effectively without imparting noticeable distortion in the image as compared to other reported techniques. 1. Introduction Computed tomography (CT) is an imaging technique capa- ble of generating high-resolution three-dimensional (3D) images of an object from two-dimensional (2D) X-ray projection data or 2D slices from 1D projection data [1, 2]. Currently, CT devices are widely used to evaluate bone specimens [3], for analysis of coronary artery walls and cancer research [4]. It is also used to screen genetically engineered small animals to investigate new drugs or therapy [5]. Nonmedical researchers have applied this technology to nondestructive testing. Modern high resolution CT machines are nowadays equipped with digital X-ray detectors (e.g., CMOS flat-panel detectors (CMOS-FPDs), or CCDs) which often create ring artifacts in the reconstructed CT images because of defective and/or miscalibrated detector elements, and dusty or damaged scintillator screens. These artifacts appear in μ-CT, C-Arm CT, and modern dental CT images in the form of a circle, centered at the center of rotation of the system. They severely impair visualization and quantification of anatomic and pathological features in the regions of interest. Therefore, removal or at least significant reduction of these artifacts is essential. Severity of the ring artifact problem can be somewhat reduced by using X-ray image intensifiers (XRII) instead of FPDs. However, low signal-to-noise ratio (SNR) and nonhomogeneous image quality across the output image due to the convex input screen of XRIIs limit their use. Therefore, C-Arm and μ-CTs are still commonly equipped with FPDs because of their several advantages over the XRIIs. Despite their high sensitivity, these detectors, however, suffer from pixel nonuniformity. Since the nonuniformity is often fluctuating depending on the operating conditions (e.g., tube voltage and tube current), it has been very difficult