Prior-Guided Metal Artifact Reduction for Iterative X-Ray Computed Tomography

Convergent-Diffusion Denoising Model for multi-scenario CT Image Reconstruction

A generic and versatile CT Image Reconstruction (CTIR) scheme can efficiently mitigate imaging noise resulting from inherent physical limitations, substantially bolstering the dependability of CT imaging diagnostics across a wider spectrum of patient cases. Current CTIR techniques often concentrate on distinct areas such as Low-Dose CT denoising (LDCTD), Sparse-View CT reconstruction (SVCTR), and Metal Artifact Reduction (MAR). Nevertheless, due to the intricate nature of multi-scenario CTIR, these techniques frequently narrow their focus to specific tasks, resulting in limited generalization capabilities for diverse scenarios. We propose a novel Convergent-Diffusion Denoising Model (CDDM) for multi-scenario CTIR, which utilizes a stepwise denoising process to converge toward an imaging-noise-free image with high generalization. CDDM uses a diffusion-based process based on a priori decay distribution to steadily correct imaging noise, thus avoiding the overfitting of individual samples. Within CDDM, a domain-correlated sampling network (DS-Net) provides an innovative sinogram-guided noise prediction scheme to leverage both image and sinogram (i.e., dual-domain) information. DS-Net analyzes the correlation of the dual-domain representations for sampling the noise distribution, introducing sinogram semantics to avoid secondary artifacts. Experimental results validate the practical applicability of our scheme across various CTIR scenarios, including LDCTD, MAR, and SVCTR, with the support of sinogram knowledge.

Metal artifact reduction method based on single spectral CT (MARSS)

BACKGROUND

For patients with metal implants, computed tomography (CT) imaging results suffer from metal artifacts, which seriously affect image evaluation and even lead to misdiagnosis. Because spectral CT technology has the advantage of quantitative imaging, basis material decomposition, and so on, the current metal artifact reduction methods are utilizing spectral information to reduce metal artifacts with good results. However, they usually require projection data from multiple spectra or energy-windows, which is difficult to realize in conventional CT.

PURPOSE

To satisfy the status quo, the aim of this work is to propose a metal artifact reduction (MAR) method based on single spectral CT (MARSS). By incorporating prior information, the average density of some base materials, and a constrained image reconstruction model is established. It forces the solution spaces of the materials to be discrete and finite, making the model easier to solve.

METHODS

The MARSS method uses the idea of discrete tomography to establish a constrained reconstruction model. By incorporating priori knowledge, the constraint forces the solution spaces of some materials to be discrete, which can effectively downsize the solution space and reduce the ill-posedness of this problem. Then, an iteration algorithm is developed to solve this model. This algorithm iterates alternately between reconstruction and discretization. It ensures that the solution spaces are discrete while the polychromatic projection of the reconstructed image converges to that of the scanned object.

RESULTS

The MRASS method significantly reduces artifacts and restores structures near metal to a large extent. Unlike the comparison MAR methods, it effectively prevents the introduction of new artifacts and distortion of the structure.

CONCLUSIONS

The MARSS method can achieve MAR based on single spectral CT. Subjective and quantitative evaluation of the results show that the method significantly improves image quality compared to competing methods.

Metal implant segmentation in CT images based on diffusion model

BACKGROUND

Computed tomography (CT) is widely in clinics and is affected by metal implants. Metal segmentation is crucial for metal artifact correction, and the common threshold method often fails to accurately segment metals.

PURPOSE

This study aims to segment metal implants in CT images using a diffusion model and further validate it with clinical artifact images and phantom images of known size.

METHODS

A retrospective study was conducted on 100 patients who received radiation therapy without metal artifacts, and simulated artifact data were generated using publicly available mask data. The study utilized 11,280 slices for training and verification, and 2,820 slices for testing. Metal mask segmentation was performed using DiffSeg, a diffusion model incorporating conditional dynamic coding and a global frequency parser (GFParser). Conditional dynamic coding fuses the current segmentation mask and prior images at multiple scales, while GFParser helps eliminate high-frequency noise in the mask. Clinical artifact images and phantom images are also used for model validation.

RESULTS

Compared with the ground truth, the accuracy of DiffSeg for metal segmentation of simulated data was 97.89% and that of DSC was 95.45%. The mask shape obtained by threshold segmentation covered the ground truth and DSCs were 82.92% and 84.19% for threshold segmentation based on 2500 HU and 3000 HU. Evaluation metrics and visualization results show that DiffSeg performs better than other classical deep learning networks, especially for clinical CT, artifact data, and phantom data.

CONCLUSION

DiffSeg efficiently and robustly segments metal masks in artifact data with conditional dynamic coding and GFParser. Future work will involve embedding the metal segmentation model in metal artifact reduction to improve the reduction effect.

Quad-Net: Quad-Domain Network for CT Metal Artifact Reduction

Metal implants and other high-density objects in patients introduce severe streaking artifacts in CT images, compromising image quality and diagnostic performance. Although various methods were developed for CT metal artifact reduction over the past decades, including the latest dual-domain deep networks, remaining metal artifacts are still clinically challenging in many cases. Here we extend the state-of-the-art dual-domain deep network approach into a quad-domain counterpart so that all the features in the sinogram, image, and their corresponding Fourier domains are synergized to eliminate metal artifacts optimally without compromising structural subtleties. Our proposed quad-domain network for MAR, referred to as Quad-Net, takes little additional computational cost since the Fourier transform is highly efficient, and works across the four receptive fields to learn both global and local features as well as their relations. Specifically, we first design a Sinogram-Fourier Restoration Network (SFR-Net) in the sinogram domain and its Fourier space to faithfully inpaint metal-corrupted traces. Then, we couple SFR-Net with an Image-Fourier Refinement Network (IFR-Net) which takes both an image and its Fourier spectrum to improve a CT image reconstructed from the SFR-Net output using cross-domain contextual information. Quad-Net is trained on clinical datasets to minimize a composite loss function. Quad-Net does not require precise metal masks, which is of great importance in clinical practice. Our experimental results demonstrate the superiority of Quad-Net over the state-of-the-art MAR methods quantitatively, visually, and statistically. The Quad-Net code is publicly available at https://github.com/longzilicart/Quad-Net.

InDuDoNet+: A deep unfolding dual domain network for metal artifact reduction in CT images

During the computed tomography (CT) imaging process, metallic implants within patients often cause harmful artifacts, which adversely degrade the visual quality of reconstructed CT images and negatively affect the subsequent clinical diagnosis. For the metal artifact reduction (MAR) task, current deep learning based methods have achieved promising performance. However, most of them share two main common limitations: (1) the CT physical imaging geometry constraint is not comprehensively incorporated into deep network structures; (2) the entire framework has weak interpretability for the specific MAR task; hence, the role of each network module is difficult to be evaluated. To alleviate these issues, in the paper, we construct a novel deep unfolding dual domain network, termed InDuDoNet+, into which CT imaging process is finely embedded. Concretely, we derive a joint spatial and Radon domain reconstruction model and propose an optimization algorithm with only simple operators for solving it. By unfolding the iterative steps involved in the proposed algorithm into the corresponding network modules, we easily build the InDuDoNet+ with clear interpretability. Furthermore, we analyze the CT values among different tissues, and merge the prior observations into a prior network for our InDuDoNet+, which significantly improve its generalization performance. Comprehensive experiments on synthesized data and clinical data substantiate the superiority of the proposed methods as well as the superior generalization performance beyond the current state-of-the-art (SOTA) MAR methods. Code is available at https://github.com/hongwang01/InDuDoNet_plus.

Virtual monochromatic imaging with projection-based material decomposition algorithm for metal artifacts reduction in photon-counting detector computed tomography

Metal artifacts present a major challenge to computed tomography (CT) because they reduce the image quality in medical diagnosis and treatment. Several metal artifact reduction (MAR) methods have been proposed to address this issue in previous studies. This study aimed to synthesize a virtual monochromatic image for MAR in CT images using projection-based material decomposition (MD) algorithms. We developed a spectral micro-CT prototype system equipped with a photon-counting detector (PCD) and PCD-CT imaging simulator to assess the performances of different MAR methods. Two projection-based MD algorithms were implemented and evaluated for their MAR performances in CT images and compared with conventional sinogram inpainting MAR methods. Different parts of digital 4D-extended cardiac torso (XCAT) phantoms with metal implants were designed to simulate various real scenarios. A homemade metal artifact evaluation (MAE) phantom was used to evaluate the MAR performance in experiments. The simulated results of the XCAT phantom indicated that the projection-based virtual monochromatic CT (VMCT) images provided better image quality than the conventional MAR images without blurring the normal tissues at the position of the metal artifacts. Various quantitative indicators support this conclusion. Additionally, the experimental results of the MAE phantom reveal that projection-based VMCT images can avoid image distortion caused by metal artifacts, unlike conventional MAR methods. In regards to the projection-based VMCT images, the simulated and experimental results demonstrated that using the linear maximum likelihood estimators with an error correction look-up table algorithm yielded better MAR performance compared to that obtained using a polynomial algorithm. Furthermore, projection-based VMCT images can not only reduce metal artifacts effectively but also simultaneously prevents object blurring at the metal artifact position and image distortion of the metal implants. Hence, the CT image quality can be further improved to increase the abilities for both preoperative and postoperative assessment of metal implants.

DICDNet: Deep Interpretable Convolutional Dictionary Network for Metal Artifact Reduction in CT Images

Computed tomography (CT) images are often impaired by unfavorable artifacts caused by metallic implants within patients, which would adversely affect the subsequent clinical diagnosis and treatment. Although the existing deep-learning-based approaches have achieved promising success on metal artifact reduction (MAR) for CT images, most of them treated the task as a general image restoration problem and utilized off-the-shelf network modules for image quality enhancement. Hence, such frameworks always suffer from lack of sufficient model interpretability for the specific task. Besides, the existing MAR techniques largely neglect the intrinsic prior knowledge underlying metal-corrupted CT images which is beneficial for the MAR performance improvement. In this paper, we specifically propose a deep interpretable convolutional dictionary network (DICDNet) for the MAR task. Particularly, we first explore that the metal artifacts always present non-local streaking and star-shape patterns in CT images. Based on such observations, a convolutional dictionary model is deployed to encode the metal artifacts. To solve the model, we propose a novel optimization algorithm based on the proximal gradient technique. With only simple operators, the iterative steps of the proposed algorithm can be easily unfolded into corresponding network modules with specific physical meanings. Comprehensive experiments on synthesized and clinical datasets substantiate the effectiveness of the proposed DICDNet as well as its superior interpretability, compared to current state-of-the-art MAR methods. Code is available at https://github.com/hongwang01/DICDNet.

Projection-domain iteration to estimate unreliable measurements

Due to the beam-hardening effect of the broad energy spectrum of the X-ray source in computed tomography, the reconstructed images usually suffer from severe artifacts when metallic objects are being imaged. Metal artifact correction methods are usually sophisticated and not practical, especially in some non-medical applications, in which the linear attenuation coefficients are unknown. This paper suggests a simple and effective algorithm to estimate the unreliable measurements. The proposed algorithm is an iterative algorithm, in which the iteration is performed in the projection domain, while the objective function is set up in the image domain. The final image is reconstructed with the conventional filtered backprojection algorithm. The feasibility of the proposed method is verified with airport bags that contain some unknown metals.

Superiorized method for metal artifact reduction

PURPOSE

Metal artifact reduction (MAR) is a challenging problem in computed tomography (CT) imaging. A popular class of MAR methods replace sinogram measurements that are corrupted by metal with artificial data, typically generated from some combination of interpolation along with other heuristics. While these "projection completion" approaches are successful in eliminating severe artifacts, secondary artifacts may be introduced by the artificial data. In this paper, we propose an approach which uses projection completion to generate a prior image, which is then incorporated into an iterative reconstruction algorithm based on the superiorization framework. The rationale is that the image produced by the iterative algorithm can inherit the desirable properties of the prior image, while also reducing secondary artifacts.

METHODS

The prior image is reconstructed using normalized metal artifact reduction (NMAR), a popular projection completion approach. The iterative algorithm is a modified version of the simultaneous algebraic reconstruction technique (SART), which reduces artifacts by incorporating a polyenergetic forward model, least-squares weighting, and superiorization. The penalty function used for superiorization is a weighted average between a total variation (TV) term and a term promoting similarity with the prior image, similar to penalty functions used in prior image constrained compressive sensing (PICCS). Because the prior is largely free of severe metal artifacts, these artifacts are discouraged from arising during iterative reconstruction; additionally, because the iterative approach uses the original projection data, it is able to recover information that is lost during the NMAR process.

RESULTS

We perform numerical experiments modeling a simple geometric object, as well as several more realistic scenarios such as metal pins, bilateral hip implants, and dental fillings placed within an anatomical phantom. The proposed iterative algorithm is largely successful at eliminating severe metal artifacts as well as secondary artifacts introduced by the NMAR process, especially lost edges of bone structures in the neighborhood of the metal regions. In one case modeling severe photon starvation, the NMAR algorithm is found to provide better results.

CONCLUSION

The proposed algorithm is effective in applying the superiorization methodology to the problem of MAR, providing better results than both NMAR and a purely total variation-based superiorization approach in nearly all cases.