Reducing Metal Artifacts in Clinical X-Ray CT via Image Reconstruction Techniques

通过图像重建技术减少临床 X 射线 CT 中的金属伪影

• 批准号: 10330750
• 负责人: Gengsheng Zeng
• 金额: $ 37.38万
• 依托单位: UTAH VALLEY UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-05-01 至 2025-05-05
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10330750
• 关键词: Address; Algorithms; Area; Back; Biomedical Engineering; Cancer Detection; Categories; Characteristics; Clinical; Clinical Data; Computed Tomography Scanners; Computer software; Contracts; Data; Data Set; Dental; Development; Diagnosis; Dose; Environment; Evaluation; Generations; Grant; Head and Neck Cancer; Health Benefit; Healthcare; Heart Diseases; Image; Implant; Industry; Internships; Learning; Machine Learning; Magnetic Resonance Imaging; Measurement; Measures; Medical Imaging; Metals; Methodology; Methods; Modeling; Morphologic artifacts; Noise; Orthopedics; Paper; Patients; Photons; Preparation; Preventive Medicine; Procedures; Protocols documentation; Publishing; Radiation; Radiation Dose Unit; Radiation therapy; Research; Research Assistant; Roentgen Rays; Scanning; Starvation; Students; Supervision; Surgical Clips; Techniques; Tissues; Training; Tube; United States; Universities; Update; Utah; Variant; Vendor; Work; X-Ray Computed Tomography; accurate diagnosis; algorithm development; base; career; clinical diagnosis; clinical examination; combat; cost effective; data acquisition; data exchange; data format; design; detector; diagnostic value; disease diagnosis; experience; feature extraction; image reconstruction; imaging modality; innovation; interest; low dose computed tomography; mathematical model; metallicity; neural network; professor; reconstruction; spectral energy; spectrograph; statistics; tool; university student

Title: Reducing Metal Artifacts in Clinical X-Ray CT via Image Reconstruction Techniques Abstract: Metallic implants such as dental fillings, surgical clips, coils, wires, and orthopedic hardware inside the patient body are very helpful in providing patients with better health benefits. On the other hand, they may cause artifacts in many medical imaging modalities such as in x-ray CT and MRI. Even though significant advances in both hardware and software have been made over the years, metal artifacts in x-ray CT scans are still troublesome. The metal artifacts appear as dark shadows and bright streaks. These artifacts, if not corrected, can severely degrade the image quality and decrease the diagnostic value of the clinical examination. X-ray generation in an x-ray tube is very inefficient; the efficiency is much less than 1%. A high photon count requirement results in the acceptance of a wide energy spectrum, which is contributed by both Bremsstrahlung and characteristic radiation. Metallic objects inside the patient body can cause photon starvation and beam hardening. These effects are nonlinear and difficult to establish an exact mathematical model for them. One of the state-of-the-art remedies is to use the dual-energy CT to measure two sets of projections, and then by using a mathematical model and combining these two data sets to estimate a set of synthetic monoenergetic x-ray measurements. Another one of the remedies is the use of a metal artifact reduction (MAR) algorithm that replaces the corrupted projection measurements in the detector with interpolation from neighboring uncorrupted projections. In data starvation situations, the dual-energy methods are not effective. The current MAR algorithms are still immature, and they may create new artifacts while trying to suppress the metal artifacts. This is a renewal application of our previous R15 grant entitled “Fast and Robust Low-Dose X-Ray CT Image Reconstruction,” in which we have successfully developed image reconstruction algorithms to combat noise. Some of our ideas formed during the last three years can be further advanced into new ideas for metal artifact reduction in x-ray CT. In this R15 renewal, we will focus on new metal artifact reduction (MAR) algorithm development based on conventional single-energy data acquisition. The innovation is the new way to set up the objective functions for optimization. The uniqueness of our objective functions is that they do not have a data fidelity term; they only contain the Bayesian terms. The Bayesian terms are formed from the metal artifact features. A gradient descent algorithm is used to optimize the objective functions and a new set of un-corrupted measurements are estimated. The final image is reconstructed by the filtered backprojection (FBP) algorithm. The proposed algorithms are cost-effective. We hypothesize that the proposed methods will be more effective than the state-of-the-art MAR methods available for commercial CT scanners. This hypothesis will be carefully evaluated in this renewed R15. Our clinical collaborators at University of Utah HealthCare will work with us closely by providing us clinical data and professional evaluation advice. This R15 project provides Utah Valley University (UVU) students with hands-on opportunities and experiences of performing real-world research in the field of healthcare. It will stimulate the interests of students so that they consider a career in biomedical and bioengineering field/industry.

题目：通过图像重建减少临床x线CT中的金属伪影

项目成果

Estimation of the Initial Image's Contributions to the Iterative Landweber Reconstruction.

• DOI: 10.1109/trpms.2017.2774834
• 发表时间: 2018-01
• 期刊: IEEE transactions on radiation and plasma medical sciences
• 影响因子: 4.4
• 作者: Zeng GL

Development of a Solvability Map.

• DOI: 10.18103/mra.v10i11.3312
• 发表时间: 2023
• 期刊: Medical research archives
• 作者: Zeng, Gengsheng L

Sparse-view tomography via displacement function interpolation.

通过位移函数插值进行稀疏视图断层扫描。

• DOI: 10.1186/s42492-019-0024-7
• 发表时间: 2019
• 期刊: Visual computing for industry, biomedicine, and art
• 作者: Zeng,GengshengL

A projection-domain iterative algorithm for metal artifact reduction by minimizing the total-variation norm and the negative-pixel energy.

• DOI: 10.1186/s42492-021-00094-w
• 发表时间: 2022-01-02
• 期刊: Visual computing for industry, biomedicine, and art
• 作者: Zeng GL

A back-projection-and-filtering-like (BPF-like) reconstruction method with the deep learning filtration from listmode data in TOF-PET.

• DOI: 10.1002/mp.15520
• 发表时间: 2022-04
• 期刊: MEDICAL PHYSICS
• 影响因子: 3.8
• 作者: Lv, Li;Zeng, Gengsheng L.;Zan, Yunlong;Hong, Xiang;Guo, Minghao;Chen, Gaoyu;Tao, Weijie;Ding, Wenxiang;Huang, Qiu
• 通讯作者: Huang, Qiu