1
|
Su T, Sun X, Yang J, Mi D, Zhang Y, Wu H, Fang S, Chen Y, Zheng H, Liang D, Ge Y. DIRECT-Net: A unified mutual-domain material decomposition network for quantitative dual-energy CT imaging. Med Phys 2021; 49:917-934. [PMID: 34935146 DOI: 10.1002/mp.15413] [Citation(s) in RCA: 15] [Impact Index Per Article: 5.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/18/2021] [Revised: 11/23/2021] [Accepted: 12/08/2021] [Indexed: 01/13/2023] Open
Abstract
PURPOSE The purpose of this paper is to present an end-to-end deep convolutional neural network to improve the dual-energy CT (DECT) material decomposition performance. METHODS In this study, we proposes a unified mutual-domain (sinogram domain and CT domain) material decomposition network (DIRECT-Net) for DECT imaging. By design, the DIRECT-Net has immediate access to mutual-domain data, and utilizes stacked convolution neural network layers for noise reduction and material decomposition. The training data are numerically generated following the fundamental DECT imaging physics. Numerical simulation of the XCAT digital phantom, experiments of a biological specimen, a calcium chloride phantom and an iodine solution phantom are carried out to evaluate the performance of DIRECT-Net. Comparisons are performed with different DECT decomposition algorithms. RESULTS Results demonstrate that the proposed DIRECT-Net can generate water and bone basis images with less artifacts compared to the other decomposition methods. Additionally, the quantification errors of the calcium chloride (75-375 mg/cm3 ) and the iodine (2-20 mg/cm3 ) are less than 4%. CONCLUSIONS An end-to-end material decomposition network is proposed for quantitative DECT imaging. The qualitative and quantitative results demonstrate that this new DIRECT-Net has promising benefits in improving the DECT image quality.
Collapse
Affiliation(s)
- Ting Su
- Research Center for Medical Artificial Intelligence, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
| | - Xindong Sun
- School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, China
| | - Jiecheng Yang
- Research Center for Medical Artificial Intelligence, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
| | - Donghua Mi
- Department of Vascular Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China
| | - Yikun Zhang
- School of Computer Science and Engineering, Southeast University, Nanjing, China
| | - Haodi Wu
- Wuhan National Laboratory for Optoelectronics (WNLO), Huazhong University of Science and Technology (HUST), Wuhan, China
| | - Shibo Fang
- Research Center for Medical Artificial Intelligence, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
| | - Yang Chen
- School of Computer Science and Engineering, Southeast University, Nanjing, China
| | - Hairong Zheng
- Paul C Lauterbur Research Center for Biomedical Imaging, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
| | - Dong Liang
- Research Center for Medical Artificial Intelligence, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China.,Paul C Lauterbur Research Center for Biomedical Imaging, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
| | - Yongshuai Ge
- Research Center for Medical Artificial Intelligence, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China.,Paul C Lauterbur Research Center for Biomedical Imaging, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
| |
Collapse
|
2
|
Lobos RA, Ghani MU, Karl WC, Leahy RM, Haldar JP. Autoregression and Structured Low-Rank Modeling of Sinogram Neighborhoods. IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING 2021; 7:1044-1054. [PMID: 35059472 PMCID: PMC8769528 DOI: 10.1109/tci.2021.3114994] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 06/14/2023]
Abstract
Sinograms are commonly used to represent the raw data from tomographic imaging experiments. Although it is already well-known that sinograms posess some amount of redundancy, in this work, we present novel theory suggesting that sinograms will often possess substantial additional redundancies that have not been explicitly exploited by previous methods. Specifically, we derive that sinograms will often satisfy multiple simple data-dependent autoregression relationships. This kind of autoregressive structure enables missing/degraded sinogram samples to be linearly predicted using a simple shift-invariant linear combination of neighboring samples. Our theory also further implies that if sinogram samples are assembled into a structured Hankel/Toeplitz matrix, then the matrix will be expected to have low-rank characteristics. As a result, sinogram restoration problems can be formulated as structured low-rank matrix recovery problems. Illustrations of this approach are provided using several different (real and simulated) X-ray imaging datasets, including comparisons against a state-of-the-art deep learning approach. Results suggest that structured low-rank matrix methods for sinogram recovery can have comparable performance to state-of-the-art approaches. Although our evaluation focuses on competitive comparisons against other approaches, we believe that autoregressive constraints are actually complementary to existing approaches with strong potential synergies.
Collapse
Affiliation(s)
- Rodrigo A Lobos
- Signal and Image Processing Institute, Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA 90089 USA
| | - Muhammad Usman Ghani
- Department of Electrical and Computer Engineering, Boston University, Boston, MA 02215 USA
| | - W Clem Karl
- Department of Electrical and Computer Engineering, Boston University, Boston, MA 02215 USA
| | - Richard M Leahy
- Signal and Image Processing Institute, Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA 90089 USA
| | - Justin P Haldar
- Signal and Image Processing Institute, Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA 90089 USA
| |
Collapse
|
3
|
Leuschner J, Schmidt M, Baguer DO, Maass P. LoDoPaB-CT, a benchmark dataset for low-dose computed tomography reconstruction. Sci Data 2021; 8:109. [PMID: 33863917 PMCID: PMC8052416 DOI: 10.1038/s41597-021-00893-z] [Citation(s) in RCA: 11] [Impact Index Per Article: 3.7] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/12/2020] [Accepted: 03/04/2021] [Indexed: 11/28/2022] Open
Abstract
Deep learning approaches for tomographic image reconstruction have become very effective and have been demonstrated to be competitive in the field. Comparing these approaches is a challenging task as they rely to a great extent on the data and setup used for training. With the Low-Dose Parallel Beam (LoDoPaB)-CT dataset, we provide a comprehensive, open-access database of computed tomography images and simulated low photon count measurements. It is suitable for training and comparing deep learning methods as well as classical reconstruction approaches. The dataset contains over 40000 scan slices from around 800 patients selected from the LIDC/IDRI database. The data selection and simulation setup are described in detail, and the generating script is publicly accessible. In addition, we provide a Python library for simplified access to the dataset and an online reconstruction challenge. Furthermore, the dataset can also be used for transfer learning as well as sparse and limited-angle reconstruction scenarios. Measurement(s) | Low Dose Computed Tomography of the Chest • feature extraction objective | Technology Type(s) | digital curation • image processing technique | Sample Characteristic - Organism | Homo sapiens |
Machine-accessible metadata file describing the reported data: 10.6084/m9.figshare.13526360
Collapse
Affiliation(s)
- Johannes Leuschner
- Center for Industrial Mathematics, University of Bremen, Bibliothekstr. 5, 28359, Bremen, Germany.
| | - Maximilian Schmidt
- Center for Industrial Mathematics, University of Bremen, Bibliothekstr. 5, 28359, Bremen, Germany.
| | - Daniel Otero Baguer
- Center for Industrial Mathematics, University of Bremen, Bibliothekstr. 5, 28359, Bremen, Germany
| | - Peter Maass
- Center for Industrial Mathematics, University of Bremen, Bibliothekstr. 5, 28359, Bremen, Germany
| |
Collapse
|
4
|
Fahimian BP, Zhao Y, Huang Z, Fung R, Mao Y, Zhu C, Khatonabadi M, DeMarco JJ, Osher SJ, McNitt-Gray MF, Miao J. Radiation dose reduction in medical x-ray CT via Fourier-based iterative reconstruction. Med Phys 2013; 40:031914. [PMID: 23464329 DOI: 10.1118/1.4791644] [Citation(s) in RCA: 28] [Impact Index Per Article: 2.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 01/12/2023] Open
Abstract
PURPOSE A Fourier-based iterative reconstruction technique, termed Equally Sloped Tomography (EST), is developed in conjunction with advanced mathematical regularization to investigate radiation dose reduction in x-ray CT. The method is experimentally implemented on fan-beam CT and evaluated as a function of imaging dose on a series of image quality phantoms and anonymous pediatric patient data sets. Numerical simulation experiments are also performed to explore the extension of EST to helical cone-beam geometry. METHODS EST is a Fourier based iterative algorithm, which iterates back and forth between real and Fourier space utilizing the algebraically exact pseudopolar fast Fourier transform (PPFFT). In each iteration, physical constraints and mathematical regularization are applied in real space, while the measured data are enforced in Fourier space. The algorithm is automatically terminated when a proposed termination criterion is met. Experimentally, fan-beam projections were acquired by the Siemens z-flying focal spot technology, and subsequently interleaved and rebinned to a pseudopolar grid. Image quality phantoms were scanned at systematically varied mAs settings, reconstructed by EST and conventional reconstruction methods such as filtered back projection (FBP), and quantified using metrics including resolution, signal-to-noise ratios (SNRs), and contrast-to-noise ratios (CNRs). Pediatric data sets were reconstructed at their original acquisition settings and additionally simulated to lower dose settings for comparison and evaluation of the potential for radiation dose reduction. Numerical experiments were conducted to quantify EST and other iterative methods in terms of image quality and computation time. The extension of EST to helical cone-beam CT was implemented by using the advanced single-slice rebinning (ASSR) method. RESULTS Based on the phantom and pediatric patient fan-beam CT data, it is demonstrated that EST reconstructions with the lowest scanner flux setting of 39 mAs produce comparable image quality, resolution, and contrast relative to FBP with the 140 mAs flux setting. Compared to the algebraic reconstruction technique and the expectation maximization statistical reconstruction algorithm, a significant reduction in computation time is achieved with EST. Finally, numerical experiments on helical cone-beam CT data suggest that the combination of EST and ASSR produces reconstructions with higher image quality and lower noise than the Feldkamp Davis and Kress (FDK) method and the conventional ASSR approach. CONCLUSIONS A Fourier-based iterative method has been applied to the reconstruction of fan-bean CT data with reduced x-ray fluence. This method incorporates advantageous features in both real and Fourier space iterative schemes: using a fast and algebraically exact method to calculate forward projection, enforcing the measured data in Fourier space, and applying physical constraints and flexible regularization in real space. Our results suggest that EST can be utilized for radiation dose reduction in x-ray CT via the readily implementable technique of lowering mAs settings. Numerical experiments further indicate that EST requires less computation time than several other iterative algorithms and can, in principle, be extended to helical cone-beam geometry in combination with the ASSR method.
Collapse
Affiliation(s)
- Benjamin P Fahimian
- Department of Radiation Oncology, Stanford University, Stanford, California 94305, USA
| | | | | | | | | | | | | | | | | | | | | |
Collapse
|
5
|
Karbeyaz BU, Naidu RC, Ying Z, Simanovsky SB, Hirsch MW, Schafer DA, Crawford CR. Variable pitch reconstruction using John's equation. IEEE TRANSACTIONS ON MEDICAL IMAGING 2008; 27:897-906. [PMID: 18599395 DOI: 10.1109/tmi.2008.922689] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 05/26/2023]
Abstract
We present an algorithm to reconstruct helical cone beam computed tomography (CT) data acquired at variable pitch. The algorithm extracts a halfscan segment of projections using an extended version of the advanced single slice rebinning (ASSR) algorithm. ASSR rebins constant pitch cone beam data to fan beam projections that approximately lie on a plane that is tilted to optimally fit the source helix. For variable pitch, the error between the tilted plane chosen by ASSR and the source helix increases, resulting in increased image artifacts. To reduce the artifacts, we choose a reconstruction plane, which is tilted and shifted relative to the source trajectory. We then correct rebinned fan beam data using John's equation to virtually move the source into the tilted and shifted reconstruction plane. Results obtained from simulated phantom images and scanner images demonstrate the applicability of the proposed algorithm.
Collapse
|
6
|
Abstract
This paper concerns image reconstruction for helical x-ray transmission tomography (CT) with multi-row detectors. We introduce two approximate cone-beam (CB) filtered-backprojection (FBP) algorithms of the Feldkamp type, obtained by extending to three dimensions (3D) two recently proposed exact FBP algorithms for 2D fan-beam reconstruction. The new algorithms are similar to the standard Feldkamp-type FBP for helical CT. In particular, they can reconstruct each transaxial slice from data acquired along an arbitrary segment of helix, thereby efficiently exploiting the available data. In contrast to the standard Feldkamp-type algorithm, however, the redundancy weight is applied after filtering, allowing a more efficient numerical implementation. To partially alleviate the CB artefacts, which increase with increasing values of the helical pitch, a frequency-mixing method is proposed. This method reconstructs the high frequency components of the image using the longest possible segment of helix, whereas the low frequencies are reconstructed using a minimal, short-scan, segment of helix to minimize CB artefacts. The performance of the algorithms is illustrated using simulated data.
Collapse
Affiliation(s)
- Hiroyuki Kudo
- Department of Computer Science, Graduate School of Systems and Information Engineering, University of Tsukuba, Japan
| | | | | | | |
Collapse
|
7
|
Noo F, Defrise M, Kudo H. General reconstruction theory for multislice X-ray computed tomography with a gantry tilt. IEEE TRANSACTIONS ON MEDICAL IMAGING 2004; 23:1109-1116. [PMID: 15377120 DOI: 10.1109/tmi.2004.829337] [Citation(s) in RCA: 14] [Impact Index Per Article: 0.7] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 05/24/2023]
Abstract
This paper discusses image reconstruction with a tilted gantry in multislice computed tomography (CT) with helical (spiral) data acquisition. The reconstruction problem with gantry tilt is shown to be transformable into the problem of reconstructing a virtual object from multislice CT data with no gantry tilt, for which various algorithms exist in the literature. The virtual object is related to the real object by a simple affine transformation that transforms the tilted helical trajectory of the X-ray source into a nontilted helix, and the real object can be computed from the virtual object using one-dimensional interpolation. However, the interpolation may be skipped since the reconstruction of the virtual object on a Cartesian grid provides directly nondistorted images of the real object on slices parallel to the tilted plane of the gantry. The theory is first presented without any specification of the detector geometry, then applied to the curved detector geometry of third-generation CT scanners with the use of Katsevich's formula for example. Results from computer-simulated data of the FORBILD thorax phantom are given in support of the theory.
Collapse
Affiliation(s)
- Frédéric Noo
- Department of Radiology, University of Utah, Salt Lake City, UT 84108, USA.
| | | | | |
Collapse
|
8
|
Manzke R, Grass M, Nielsen T, Shechter G, Hawkes D. Adaptive temporal resolution optimization in helical cardiac cone beam CT reconstruction. Med Phys 2003; 30:3072-80. [PMID: 14713073 DOI: 10.1118/1.1624756] [Citation(s) in RCA: 80] [Impact Index Per Article: 3.8] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/07/2022] Open
Abstract
Cone beam computed tomography scanners in combination with heart rate adaptive reconstruction schemes have the potential to enable cardiac volumetric computed tomography (CT) imaging for a larger number of patients and applications. In this publication, an adaptive scheme for the automatic and patient-specific reconstruction optimization is introduced to improve the temporal resolution and image quality. The optimization method permits the automatic determination of the required amount of gated helical cone beam projection data for the reconstruction volume. It furthermore allows one to optimize subvolume reconstruction yielding an increased temporal resolution. In addition, methods for the assessment of the temporal resolution are given which enable a quantitative documentation of the reconstruction improvements. Results are presented for patient data sets acquired in low pitch helical mode using a 16-slice cone beam CT system with parallel ECG recording.
Collapse
Affiliation(s)
- R Manzke
- Philips Research Laboratories, Sector Technical Systems, Hamburg, Germany and Imaging Sciences Group, Guy's Hospital Campus, KCL, London, United Kingdom.
| | | | | | | | | |
Collapse
|
9
|
Chen L, Liang Y, Heuscher DJ. General surface reconstruction for cone-beam multislice spiral computed tomography. Med Phys 2003; 30:2804-21. [PMID: 14596317 DOI: 10.1118/1.1610291] [Citation(s) in RCA: 9] [Impact Index Per Article: 0.4] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/07/2022] Open
Abstract
A new family of cone-beam reconstruction algorithm, the General Surface Reconstruction (GSR), is proposed and formulated in this paper for multislice spiral computed tomography (CT) reconstructions. It provides a general framework to allow the reconstruction of planar or nonplanar surfaces on a set of rebinned short-scan parallel beam projection data. An iterative surface formation method is proposed as an example to show the possibility to form nonplanar reconstruction surfaces to minimize the adverse effect between the collected cone-beam projection data and the reconstruction surfaces. The improvement in accuracy of the nonplanar surfaces over planar surfaces in the two-dimensional approximate cone-beam reconstructions is mathematically proved and demonstrated using numerical simulations. The proposed GSR algorithm is evaluated by the computer simulation of cone-beam spiral scanning geometry and various mathematical phantoms. The results demonstrate that the GSR algorithm generates much better image quality compared to conventional multislice reconstruction algorithms. For a table speed up to 100 mm per rotation, GSR demonstrates good image quality for both the low-contrast ball phantom and thorax phantom. All other performance parameters are comparable to the single-slice 180 degrees LI (linear interpolation) algorithm, which is considered the "gold standard." GSR also achieves high computing efficiency and good temporal resolution, making it an attractive alternative for the reconstruction of next generation multislice spiral CT data.
Collapse
Affiliation(s)
- Laigao Chen
- School of Health Sciences, Purdue University, West Lafayette, Indiana 47907, USA
| | | | | |
Collapse
|