18 January 2024 in IEEE Transactions on Medical Imaging
Metal artifacts caused by the presence of metallic implants tremendously degrade the quality of reconstructed computed tomography (CT) images and therefore affect the clinical diagnosis or reduce the accuracy of organ delineation and dose calculation in radiotherapy. Although various deep learning methods have been proposed for metal artifact reduction (MAR), most of them aim to restore the corrupted sinogram within the metal trace, which removes beam hardening artifacts but ignores other components of metal artifacts. In this paper, based on the physical property of metal artifacts which is verified via Monte Carlo (MC) simulation, we propose a novel physics-inspired non-local dual-domain network (PND-Net) for MAR in CT imaging. Specifically, we design a novel non-local sinogram decomposition network (NSD-Net) to acquire the weighted artifact component and develop an image restoration network (IR-Net) to reduce the residual and secondary artifacts in the image domain. To facilitate the generalization and robustness of our method on clinical CT images, we employ a trainable fusion network (F-Net) in the artifact synthesis path to achieve unpaired learning. Furthermore, we design an internal consistency loss to ensure the data fidelity of anatomical structures in the image domain and introduce the linear interpolation sinogram as prior knowledge to guide sinogram decomposition. NSD-Net, IR-Net, and F-Net are jointly trained so that they can benefit from one another. Extensive experiments on simulation and clinical data demonstrate that our method outperforms state-of-the-art MAR methods.
26 October 2023 in IEEE Transactions on Nuclear Science
Talbot-Lau grating interferometer (GI) can provide valuable tri-contrast (absorption, differential phase contrast, and dark field) information about the samples to be imaged. However, the source grating G0 presents severe challenges associated with fabrication and limited efficiency. In this study, a novel cold-cathode flat-panel X-ray source with microarray anode target was proposed for GI computed tomography, which can eliminate the use of G0. Its cathode is characterized as densely arranged ZnO nanowires (NWs) to provide large amounts of electrons, whereas the micro periodic distributed Al-Mo-Al strips were utilized as the anode target to generate the structured X-ray illumination required for Talbot-Lau GI. The performance of the source was fully studied by accurately modeling its geometry via EGSnrc. The flat-panel X-ray source GI computed tomography was then simulated by utilizing Fresnel-Kirchhoff diffraction theory. Furthermore, a PMMA cylinder and a Shepp-Logan phantom were utilized to evaluate the imaging performance of the system. Due to the elimination of G0, our proposed flat-panel source GI has 153.05% higher X-ray photons utilization efficiency than the conventional Talbot-Lau GI, which results in tri-contrast images with less noise. The proposed X-ray source can replace the lab source and grating G0 and may provide new possibilities for Talbot-Lau GI.
14 July 2025 in Optics Express
Different from the conventional X-ray point tube, the cold-cathode flat-panel X-ray source (FPXS) achieved planar integration of tens of thousands of X-ray point sources. However, the X-ray source distribution of the FPXS at the generating plane (Mo target) is unknown and cannot be directly measured, which hinders its practical use. To address this issue, we propose to decompose the X-ray distribution of the FPXS from the aliased projection of a coded mask. The aliased projection was a circular convolution of the source distribution of FPXS and the basic coded mask, which uniquely encodes the FPXS flux. Besides, the maximum-likelihood expectation-maximization (ML-EM) algorithm was derived to reconstruct the source distribution. Both Monte Carlo (MC) simulation and real experiments were conducted to demonstrate the effectiveness of the proposed method. In the real experiments, the X-ray emission shapes formed by placing lead plates on the surface of the anode substrate of two FPXSs were successfully reconstructed by the proposed method. Moreover, knowing and then compensating the non-uniformity distribution of the FPXS is paramount for future novel and practical applications of the X-ray source.