Journal Article10.1109/TCI.2015.2461492
A Model-Based Image Reconstruction Algorithm With Simultaneous Beam Hardening Correction for X-Ray CT
175
TL;DR: This paper presents a novel model-based iterative reconstruction algorithm that incorporates beam hardening correction (MBIR-BHC), which significantly reduces several reconstruction artifacts without advanced knowledge of the X-ray spectrum and material properties.
read more
Abstract: Beam hardening is a well-known effect in X-ray CT scanning that is caused by the interaction of a broad polychromatic source spectrum with energy-dependent material attenuation. If the scanned object only consists of a single material, the beam hardening effect can be corrected by sinogram precorrection techniques. However, when multiple materials are present, it becomes much more difficult to fully compensate for this distortion; in general, the beam hardening can contribute to reconstruction artifacts such as cupping and streaking. In this paper, we present a novel model-based iterative reconstruction algorithm that incorporates beam hardening correction (MBIR-BHC). Unlike most correction algorithms, which require knowledge of the X-ray spectrum or mass attenuation functions, the MBIR-BHC algorithm works by simultaneously reconstructing the image and estimating the beam-hardening function. The method is based on the assumption that the object is formed by a combination of two distinct materials that can be separated based on their densities. We formulate a poly-energetic X-ray forward model using a polynomial function of two material projections: one for the low-density material and one for the high. We then develop an alternating minimization algorithm for jointly estimating the reconstructed image, the material segmentation, and the coefficients of the two component polynomial that models the beam-hardening function. With this approach, the spectrum and mass attenuation functions are not needed in advance, and the correction is adapted to the dataset being reconstructed. We examine the performance of the proposed algorithm using both simulated and real datasets. Results indicate that the MBIR-BHC algorithm significantly reduces several reconstruction artifacts without advanced knowledge of the X-ray spectrum and material properties.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
A review on medical imaging synthesis using deep learning and its clinical applications.
TL;DR: This paper summarized the recent developments of deep learning-based methods in inter- and intra-modality image synthesis by listing and highlighting the proposed methods, study designs, and reported performances with related clinical applications on representative studies.
265
Fast Enhanced CT Metal Artifact Reduction Using Data Domain Deep Learning
TL;DR: This work treats the projection data corresponding to dense, metal objects as missing data and train an adversarial deep network to complete the missing data directly in the projection domain, resulting in an end-to-end metal artifact reduction algorithm that is computationally efficient textcolorredand therefore practical and fits well into existing CT workflows allowing easy adoption in existing scanners.
143
Dense Recurrent Neural Networks for Accelerated MRI: History-Cognizant Unrolling of Optimization Algorithms
TL;DR: The results in reconstruction of the fastMRI knee dataset show that the proposed history-cognizant approach reduces residual aliasing artifacts compared to its conventional unrolled counterpart without requiring extra computational power or increasing reconstruction time.
97
Dual-energy crystal-analyzer scheme for spectral tomography.
TL;DR: A method for adjusting a crystal analyzer to separate two characteristic lines from the spectrum of a conventional X-ray tube for simultaneous registration of tomographic projections is proposed.
87
Comparison Between Pre-Log and Post-Log Statistical Models in Ultra-Low-Dose CT Reconstruction
Lin Fu,Tzu-Cheng Lee,Soo Mee Kim,Adam M. Alessio,Paul E. Kinahan,Zhiqian Chang,Ken D. Sauer,Mannudeep K. Kalra,Bruno De Man +8 more
TL;DR: Several pre-log and post-log MBIR algorithms under a unified framework are developed and shown to achieve notably better quantitative accuracy in ultra-low-dose CT, although in less extreme cases, post- logs with handcrafted pre-processing remains a competitive alternative.
73
References
Regression Shrinkage and Selection via the Lasso
TL;DR: A new method for estimation in linear models called the lasso, which minimizes the residual sum of squares subject to the sum of the absolute value of the coefficients being less than a constant, is proposed.
Energy-selective reconstructions in X-ray computerized tomography.
TL;DR: It is shown that for any material, complete energy spectral information may be summarized by a few constants which are independent of energy, and a technique is presented which uses simple, low-resolution, energy spectrum measurements and conventional computerized tomography techniques to calculate these constants.
2.2K
•Book
Computed Tomography: Principles, Design, Artifacts, and Recent Advances
Jiang Hsieh
- 01 Jan 2003
TL;DR: Introduction Preliminaries Image Reconstruction Image Presentation Key Performance Parameters of a CT Scanner Major Components of CT scanner Image Artifacts: Appearances, Causes, and Corrections Computer Simulation and Analysis.
1.7K