Cupping artifacts correction for polychromatic X-ray cone-beam computed tomography based on projection compensation and hardening behavior

https://doi.org/10.1016/j.bspc.2019.101823Get rights and content

Highlights

  • The idea is to estimate monochromatic projections from polychromatic projections using a polynomial model and the prior knowledge on the relationship between the ray length and corresponding projection.

  • The artifact corrector is based on projection compensation and an analysis of the hardening characteristics.

  • The adaptive estimation of the tradeoff factor λ is exploited to the model for showing the characteristics and details of the correction from polychromatic to monochromatic X-ray in this work.

Abstract

Based on the beam hardening behavior of polychromatic X-ray, the aim of this paper is to address and test an improved experimental and theoretical investigations of the distorted polychromatic projection data to generate high qualified imaging by reducing the cupping artifacts and noise in Cone-Beam computed tomography (CBCT). Since the hardening information has a large relationship with the span that the ray passing through the object, a method of obtaining the lengths that the ray across the binary image is proposed. As the knowledge, a new weighted and compensated correction model is constructed to primarily eliminate cupping artifacts. Finally, studies on polychromatic projection are verified by using pieces on CBCT system. Study results show effective suppression of cupping artifacts of polychromatic projection. For experimental, the root mean square error (RMSE) of the proposed method is reduced by 31.6 %, peak signal to noise ratio (PSNR), and universal quality index (UQI) are increased by 16.8 % and 12.8 % for phantom #Bushing, respectively; which also indicate better uniformity of the results for phantom #Wheel, where the RMSE is reduced by 30.6 %, PSNR and UQI were increased by 19.4 % and 15.9 %, respectively. The proposed algorithm is anticipated to find its utility in industrial nondestructive testing wherein the image reconstruction for the complex geometry and high energy X-ray imaging.

Introduction

Cone-beam computed tomography (CBCT) plays a crucial role in both medical imaging and industrial non-destructive testing (NDT) with high efficiency and precision [1,2], which is equipped with X-ray beams that emit photons of different energy and frequency. X-ray computed tomography became an important instrument for quality assurance in industry products as a non-destructive testing tool for inspection, evaluation, analysis and dimensional metrology. Thus, a high-quality image is required. Due to the polychromatic nature of X-ray energy in X-ray computed tomography, this leads to errors in the attenuation coefficient. This leads to a distortion or blurring-like cupping and streak in the reconstruction images, where a significant decrease in imaging quality is observed [3]. However, most reconstruction approaches do not properly examine the non-linear nature of the polychromatic X-rays. When polychromatic X-ray beams used during CT imaging pass through a patient, soft and low energy X-rays [4], which are of little importance in image formation, are preferentially absorbed to a great extent compared to high-energy photons. The consequence of this selective absorption phenomenon is the increase in the patient’s absorbed dose and non-linear increase in the beam’s average energy, which is the beam hardening effect [5,6]. As a result, the total attenuation of X-rays and therefore the associated log-processed transmission data will no longer be a linear function of tissue thickness.

Since the low-energy photons are attenuated more than the high-energy ones, the more transmission, the higher the average energy of photons when interacts with objects [7,8]. The most widely used Filtered back-projection (FBP) algorithms in CT reconstruction assume a linear propagation model for the detected photons and as such fail to consider the non-linear beam hardening effect [9]. Consequently, the reconstructed images exhibit cupping artifacts and reduced CT numbers behind bony structures and streak artifacts around metallic objects [10]. These artifacts may mimic or obscure pathologic lesions, rendering the interpretation of CT images complicated and leading to equivocal findings. This leads to severe noise and other artifacts [11] and inhomogeneity which degrades image quality and makes diagnosis difficult or even impossible [[12], [13], [14]]. The poly-chromaticity of the X-ray beam therefore not only increases the patient dose but also induces beam hardening artifacts. Hence, there is a need for taking appropriate measures toward concomitant patient dose reduction and beam hardening correction.

As a focused problem to be solved, several approaches for beam hardening have been proposed, which can be classified as dual-energy methods [15,16], statistical iterative methods [[17], [18], [19], [20]] and sinogram inpainting methods [[21], [22], [23], [24]]. Furthermore, specially designed filters known as beam shapers are used in CT scanners to shape the x-ray beam to deliver dose with the appropriate spatial distribution. As a consequence of pre-filtering, both beam hardening effect and patient’s absorbed dose are intrinsically reduced, however, as a compromise, statistical noise or quantum mottle is increased, which in turn impairs image quality and low-contrast detectability. Despite some good preliminary results that have been obtained, the dual-energy correction requires lengthy post-processing and the high dose of radiation is a limitation. While, the iterative method requires prior information about the energy spectrum of the incident X-ray and energy-dependent attenuation coefficients of the materials [25,26]. Brabant et al. [27], employed an iterative reconstruction algorithm and took the effect of beam hardening into account during the reconstruction. To solve the issue of sinogram-consistency, Cao et al., [28] presented a beam hardening correction regression model to increase the dimensional measurement accuracy, which is trained with the simulated data afterward. Romano et al., [29] using a linearization procedure of the beam hardening curve applied after the reconstruction process to provide accurate results. Zhao et al., [30] developed a fast and accurate beam hardening correction method by modeling physical interactions between X-ray photons and materials for computed tomography (CT) imaging. The nonlinear attenuation process of the X-ray projection is modeled by reprojecting a template image with the estimated polychromatic spectrum. Sarkar et al., [31] utilizes discrete cosine transform-based missing data estimation and an edge-preserving smoothing filter for the suppression of beam hardening artifacts. Park et al., [32] proposes a sinogram-consistency learning method. to repair inconsistent sinogram by removing the primary metal-induced beam hardening factors along the metal trace in the sinogram in polychromatic computerized tomography (CT).

The beam hardening is modeled and incorporated in the forward projector of the simultaneous algebraic reconstruction technique (SART). Various inpainting-based artifact reduction methods had been suggested, such as interpolation [33,34], Poisson in painting [35], wavelet [36], and total variation [37]. But preliminary studies had found limited performance, including the inaccurate interpolation who had introduced additional artifacts to the image and had deteriorated morphological information. Kyriakou et al. [38] proposed a beam hardening correction which replaced the corrupted projection data by interpolating, it had effectively reduced the artifacts, but the edge information lost. The traditional method uses a wedge-shaped phantom to obtain the curve. Huang et al. [8] constructed a new fitting function with satisfactory stability of curve shape through simplifying beam hardening data based on histogram statistics of traversing lengths of the ray. Zou et al. [39] presented algorithm utilizes image segmentation to obtain the hardening object, and subsequently dilates segmented image to suppress dependence of the exactness of segmentation to correction result based on the modified TV algorithm. The previous methods are designed to reduce the higher-order artifacts induced by beam hardening in require material correction using calibration for global cupping correction.

Inspired by the process and the properties of non-linear beam hardening of different materials, we aim to address the improved experimental and theoretical investigations of the distorted polychromatic projection data to generate high qualified imaging by reducing the cupping artifacts and noise in Cone-Beam computed tomography (CBCT) in this work. The method is evaluated using CT images, and the experiments show that the proposed method is attractive as a preferred quality for polychromatic X-rays imaging. The main contributions of the approach are presented as follows:

  • 1)

    A new effective cupping artifacts correction for a polychromatic X-ray is proposed. The idea is to estimate monochromatic projections from polychromatic projections using a polynomial model and the prior knowledge on the relationship between the ray length and corresponding projection.

  • 2)

    The artifact corrector is based on projection compensation and an analysis of the hardening characteristics. To reveal the hardening effects of different materials, the introduction and adaptive estimation of the tradeoff factor λ is exploited to the model for showing the characteristics and details of the correction from polychromatic to monochromatic X-ray in this work.

The remainder of the paper is organized as follows. Section 2 provides a relevant description of notions and the approach for artifacts correction algorithm. Section 3 presents the experimental results and the performance comparison with the state-of-the-art methods. Finally, Section 4 remarks on the conclusion of this paper.

Section snippets

Cupping artifacts and correction scheme

As has been mentioned that when the polychromatic X-ray beams used during CT imaging pass through the body or the object, soft and low energy X-rays are preferentially absorbed to a great extent compared to high-energy photons. With the X-ray path increased, the proportion of high-energy photons increase and the average energy of ray arise [40]. Then, the attenuation coefficient μ is no longer a constant, which shows a decreasing curve as the path gets longer. Clearly, this phenomenon leads to

Experimental results

To avoid the issue that the ray cannot penetrate the objects especially in industry application (It has been a hot topic in CT imaging), we design a #brushing in a diameter of Φ40 with the material of iron to explain the feasibility and correctness of the proposed modeling based on the cylindrical structure. Furthermore, we design the #wheel in a diameter of Φ60 with the material of titanium and adding some auxiliary rib structure to change the path of the ray’s attenuation. As has been shown

Conclusion

In this paper, an effective cupping artifacts correction method has been described. A method of obtaining the traversing lengths between the ray and the binary image is proposed to obtain discrete beam hardening data. Based on the characteristics of hardening behavior and knowledge, a new weighted and compensated model is constructed. The performance of the proposed method has been verified on the CBCT system. The preliminary data obtained in the experimental investigation shows that the

Acknowledgments

This work was supported in part by the National Natural Science Foundation of China (51675437 and 51605389); Fund of Ministry of Industry and Information Technology of China (MJ-2017-F-05); The Fundamental Research Funds for the Central Universities (31020190504006); China Postdoctoral Science Foundation (2019M653749). The authors are grateful to the anonymous reviewers for their valuable comments.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fuqiang Yang is with the Key Laboratory of High Performance Manufacturing for Aero Engine (Northwestern Polytechnical University), Ministry of Industry and Information Technology; Engineering Research Center of Advanced Manufacturing Technology for Aero engine (Northwestern Polytechnical University), Ministry of  Education, Xi’an 710072, China.

References (49)

  • S.J. Tang et al.

    Optimization based beam-hardening correction in CT under data integral invariant constraint

    Phys. Med. Biol.

    (2018)
  • W.C. Cao et al.

    A simulation-based study on the influence of the x-ray spectrum on the performance of multi-material beam hardening correction algorithms

    Meas. Sci. Technol.

    (2018)
  • K.D. Huang et al.

    Noise suppression methods in beam hardening correction for X-ray computed tomography

  • Y.I. Nesterets et al.

    Noise propagation in x-ray phase-contrast imaging and computed tomography

    J. Phys. D Appl. Phys.

    (2014)
  • H.S. Park et al.

    Metal artifact reduction for polychromatic X-ray CT based on a beam-hardening corrector

    IEEE Trans. Med. Imaging

    (2016)
  • A. Shiras et al.

    Beam hardening correction using cone beam consistency conditions

    IEEE Trans. Med. Imaging

    (2018)
  • Z. Christoph et al.

    Decomposing a prior-CT-based cone-beam CT projection correction algorithm into scatter and beam hardening components

    Phys. Imaging Radiat. Oncol.

    (2017)
  • L. Yu et al.

    Dual-energy CT-based monochromatic imaging

    AJR Am. J. Roentgenol.

    (2012)
  • Y.B. Chang et al.

    Metal artifact reduction algorithm for single energy and dual energy CT scans

  • H. Shi et al.

    Reduce beam hardening artifacts of polychromatic X-ray computed tomography by an iterative approximation approach

    J. Xray Sci. Technol.

    (2017)
  • G.V. Gompel et al.

    Iterative correction of beam hardening artifacts in CT

    Med. Phys.

    (2011)
  • S.J. Tang et al.

    Optimization based beam-hardening correction in CT under data integral invariant constraint

    Phys. Med. Biol.

    (2018)
  • H. Wang et al.

    A new approach for reducing beam hardening artifacts in polychromatic X-ray computed tomography using more accurate prior image

    J. Xray Sci. Technol.

    (2018)
  • R. Tovey et al.

    Directional sinogram inpainting for limited angle tomography

    Inverse Probl.

    (2018)
  • Cited by (11)

    • Applications of Computed Tomography (CT) in environmental soil and plant sciences

      2023, Soil and Tillage Research
      Citation Excerpt :

      In general, the X-ray spectrum used in conventional CT scanners has a polychromatic characteristic. Most reconstruction methods cannot correctly check the nonlinearity of polychromatic X-rays (Cao et al., 2018), which causes beam hardening that can lead to artifacts and degrade the quality of the reconstructed image (Yang et al., 2020). Better peak separation is observed with the application of filters.

    • Research on drop-weight impact of continuous carbon fiber reinforced 3D printed honeycomb structure

      2021, Materials Today Communications
      Citation Excerpt :

      According to the experimental results, the bearing capacity of continuous fiber reinforced honeycomb under drop-weight impact has a certain gap with the expected value and the FE simulation results. In order to better evaluate the impact resistance of the structure, the industrial cone beam CT nondestructive testing is used to detect the internal damage of honeycomb caused by impact, the detection parameters of industrial cone beam CT are as follows [35]: DSD(Distance between source and detector)= 1222 mm, DSO(Distance between source and origin)= 872 mm, scanning angle and the number of circular projections are 360° and 720 respectively, the diameter of the source is 0.4 mm, voltage and current of the source are 100 kV and 1.2 mA respectively. The detection image is shown in Fig. 11, obvious internal stratification phenomenon appears in the structure after impact, as shown in Fig. 11(b).

    • Analyzing the microstructure of a fresh sorbet with X-ray micro-computed tomography: Sampling, acquisition, and image processing

      2021, Journal of Food Engineering
      Citation Excerpt :

      Secondary radiation is particularly intense from the inner regions of the particles; it induces an overestimation of the X-rays received by the detector and therefore an underestimation of the attenuation of X-rays by the central region of the particles. Finally, the inner region of the particles appears darker (Wils, 2011; Yang et al., 2020). The results obtained by performing the same image processing on the ice phase as for the air phase (see section 3.3.1) are shown in Fig. 5(b).

    • GPU-accelerated polyenergetic forward projection for 9 MeV industrial CT system

      2023, Insight: Non-Destructive Testing and Condition Monitoring
    View all citing articles on Scopus

    Fuqiang Yang is with the Key Laboratory of High Performance Manufacturing for Aero Engine (Northwestern Polytechnical University), Ministry of Industry and Information Technology; Engineering Research Center of Advanced Manufacturing Technology for Aero engine (Northwestern Polytechnical University), Ministry of  Education, Xi’an 710072, China.

    Dinghua Zhang is with the Key Laboratory of High Performance Manufacturing for Aero Engine (Northwestern Polytechnical University), Ministry of Industry and Information Technology; Engineering Research Center of Advanced Manufacturing Technology for Aero engine (Northwestern Polytechnical University), Ministry of  Education, Xi’an 710072, China.

    Hua Zhang is with the School of Mechanical Engineering, Northwestern Polytechnical University, Xi’an 710072, China.

    Kuidong Huang is with the Key Laboratory of High Performance Manufacturing for Aero Engine (Northwestern Polytechnical University), Ministry of Industry and Information Technology; Engineering Research Center of Advanced Manufacturing Technology for Aero engine (Northwestern Polytechnical University), Ministry of  Education, Xi’an 710072, China.

    View full text