Article
Engineering, Electrical & Electronic
Davis Gilton, Gregory Ongie, Rebecca Willett
Summary: This paper presents an alternative approach utilizing an infinite number of iterations, which consistently improves reconstruction accuracy beyond state-of-the-art alternatives; additionally, the computational budget can be chosen at test time to optimize context-dependent trade-offs between accuracy and computation.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2021)
Article
Mathematics
Yuhan Yin, Juan Liu
Summary: In this study, the problem of determining the shape of an object in two-dimensional elasticity from the far-field of a single incident wave was considered. An iterative hybrid method was applied to address this problem, eliminating the need for a forward solver and exact boundary conditions. Reconstruction algorithms were established for objects with Dirichlet, Neumann, and Robin boundary conditions, as well as for objects with unknown physical properties by introducing a general boundary condition. Numerical experiments demonstrated the effectiveness of the proposed method.
Article
Computer Science, Information Systems
Ali Imran Sandhu, Abdulla Desmal, Hakan Bagci
Summary: An efficient nonlinear contrast source inversion scheme is proposed for electromagnetic imaging of sparse two-dimensional investigation domains, tackling non-linearity using the NLW iterations and incorporating a self-adaptive projected accelerated steepest descent (A-PASD) algorithm for enhanced efficiency. The results demonstrate the accuracy, efficiency, and applicability of the proposed scheme.
Article
Computer Science, Interdisciplinary Applications
Jinxi Xiang, Yonggui Dong, Yunjie Yang
Summary: FISTA-Net is a model-based deep learning network that optimizes parameters for different imaging tasks without tuning, outperforming existing methods and exhibiting good generalization ability under different noise levels.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2021)
Article
Engineering, Electrical & Electronic
Yu Sun, Jiaming Liu, Mingyang Xie, Brendt Wohlberg, Ulugbek Kamilov
Summary: CoIL is a new deep-learning methodology for continuous representation of measurements, training a multilayer perceptron to encode complete measurement fields by mapping coordinates to responses. It is a self-supervised method that generates new measurements without the need for training examples besides those of the test object, improving the performance of various reconstruction methods.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2021)
Article
Computer Science, Artificial Intelligence
Adnan Qayyum, Inaam Ilahi, Fahad Shamshad, Farid Boussaid, Mohammed Bennamoun, Junaid Qadir
Summary: In recent years, deep learning techniques have made significant progress in solving inverse imaging problems, outperforming hand-crafted approaches. Deep learning models make use of large datasets to predict unknown solutions to the inverse problems. A new paradigm called untrained neural network prior (UNNP) has been proposed, which utilizes a single image for deep model training in various inverse tasks. This article comprehensively reviews studies on UNNP and its applications for different tasks, highlighting open research problems that require further investigation.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Engineering, Electrical & Electronic
Andre Costa Batista, Lucas S. Batista, Ricardo Adriano
Summary: This study proposes a method for solving the inverse problem of microwave imaging using quadratic programming, with the ability to enforce prior information accurately during image reconstruction. Experimental results demonstrate good performance of the method in various scenarios, particularly in reducing errors effectively and speeding up convergence.
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION
(2021)
Article
Engineering, Electrical & Electronic
Davis Gilton, Gregory Ongie, Rebecca Willett
Summary: Researchers have proposed two novel procedures that allow networks to adapt to changes in the forward model without full understanding of the change. These methods do not require access to more labeled data and have been successful in a variety of inverse problems such as deblurring, super-resolution, and undersampled image reconstruction in magnetic resonance imaging.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2021)
Article
Operations Research & Management Science
Barbara Kaltenbacher, Kha Van Huynh
Summary: In this paper, we study the formulation of inverse problems as constrained minimization problems and their iterative solution by gradient or Newton type methods. We carry out a convergence analysis in the sense of regularization methods and discuss applicability to the problem of identifying the spatially varying diffusivity in an elliptic PDE from different sets of observations. Among these is a novel hybrid imaging technology known as impedance acoustic tomography, for which we provide numerical experiments.
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS
(2022)
Article
Geochemistry & Geophysics
Yusong Wang, Zheng Zong, Siyuan He, Zhun Wei
Summary: Recently, deep learning methods have achieved significant success in solving inverse scattering problems (ISPs). However, there has been limited exploration of learning approaches that work in different spaces, such as frequency and real space, for ISPs. In this study, multiple-space deep learning schemes (MSDLSs) that incorporate both frequency-space and real-space processing are investigated. The proposed MSDLSs, including a network in the low-frequency subspace and serial and parallel MSDLSs, show consistent improvements over the traditional backpropagation scheme (BPS) in both synthetic and experimental tests. These findings suggest that the proposed MSDLSs have the potential to be applied in other inverse problems that require the incorporation of multiple-space information.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Editorial Material
Optics
Chonghang Zhao, Hanfei Yan
Summary: Deep neural network can significantly improve tomography reconstruction with limited data. A recent study combining the ptycho-tomography model with the 3D U-net demonstrated a reduction in both the number of projections and computation time, and showed its potential for integrated circuit imaging that requires high-resolution and fast measurement speed.
LIGHT-SCIENCE & APPLICATIONS
(2023)
Article
Geochemistry & Geophysics
Oktay Karakus, Alin Achim
Summary: In this article, solutions for inverse problems in SAR imaging are investigated with a proposed convex proximal splitting method that optimizes a cost function with a nonconvex Cauchy-based penalty. The performance of the penalty function is evaluated on three standard SAR imaging inverse problems, showing better image reconstruction results compared to other classical penalty functions.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Mathematics, Applied
Ruixue Gu, Bo Han, Shanshan Tong, Yong Chen
Summary: This paper proposes and analyzes a novel method for solving inverse problems in Banach spaces, by combining homotopy perturbation iteration and Kaczmarz method with uniformly convex penalty terms. The method shows convergence in the exact data case and is able to reconstruct special features of solutions such as sparsity and piecewise constancy.
JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS
(2021)
Article
Engineering, Electrical & Electronic
Andreas Asmann, Joao F. C. Mota, Brian D. Stewart, Andrew Michael Wallace
Summary: The proposed architecture reconstructs depth images from raw photon count data using compressive sensing techniques, achieving real-time high-resolution depth map reconstruction.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2022)
Article
Geochemistry & Geophysics
Xiaoniu Zhang, Zhiqin Zhao
Summary: This letter proposes a new scheme based on the contraction integral equation method to solve highly nonlinear inverse scattering problems with noise disturbance. By controlling the weight factor, the imaging quality of the inversion algorithm is improved and the convergence is enhanced without significant increase in computational costs.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Oliver Maier, Jasper Schoormans, Matthias Schloegl, Gustav J. Strijkers, Andreas Lesch, Thomas Benkert, Tobias Block, Bram F. Coolen, Kristian Bredies, Rudolf Stollberger
MAGNETIC RESONANCE IN MEDICINE
(2019)
Article
Clinical Neurology
Houchun H. Hu, Thomas Benkert, Mark Smith, Jeremy Y. Jones, Aaron S. McAllister, Jerome A. Rusin, Ramkumar Krishnamurthy, Kai Tobias Block
Article
Radiology, Nuclear Medicine & Medical Imaging
Michael P. Recht, Kai Tobias Block, Hersh Chandarana, Jennifer Friedland, Thomas Mullholland, Donal Teahan, Roy Wiggins
AMERICAN JOURNAL OF ROENTGENOLOGY
(2019)
Article
Clinical Neurology
J. L. Mogen, K. T. Block, N. K. Bansal, J. T. Patrie, S. Mukherjee, E. Zan, M. Hagiwara, G. M. Fatterpekar, S. H. Patel
AMERICAN JOURNAL OF NEURORADIOLOGY
(2019)
Article
Radiology, Nuclear Medicine & Medical Imaging
Houchun H. Hu, Thomas Benkert, Jeremy Y. Jones, Aaron S. McAllister, Jerome A. Rusin, Ramkumar Krishnamurthy, Kai Tobias Block
Article
Radiology, Nuclear Medicine & Medical Imaging
Tom Hilbert, Ding Xia, Kai Tobias Block, Zidan Yu, Riccardo Lattanzi, Daniel K. Sodickson, Tobias Kober, Martijn A. Cloos
MAGNETIC RESONANCE IN MEDICINE
(2020)
Article
Radiology, Nuclear Medicine & Medical Imaging
Azadeh Sharafi, Rahman Baboli, Marcelo Zibetti, Krishna Shanbhogue, Sonja Olsen, Tobias Block, Hersh Chandarana, Ravinder Regatte
MAGNETIC RESONANCE IN MEDICINE
(2020)
Article
Cardiac & Cardiovascular Systems
Philip M. Robson, Vittoria Vergani, Thomas Benkert, Maria Giovanna Trivieri, Nicolas A. Karakatsanis, Ronan Abgral, Marc R. Dweck, Pedro R. Moreno, Jason C. Kovacic, Kai Tobias Block, Zahi A. Fayad
Summary: This study compared the impact of using multi-tissue-class MRAC with simple two-class MRAC on cardiac PET/MR imaging. The results showed that multi-tissue-class MRAC significantly affected the quantification of 18F-fluorodeoxyglucose activity in the myocardium, but only had a moderate impact on the appearance of the PET image.
JOURNAL OF NUCLEAR CARDIOLOGY
(2021)
Article
Biophysics
Hassan Haji-Valizadeh, Li Feng, Liliana E. Ma, Daming Shen, Kai Tobias Block, Joshua D. Robinson, Michael Markl, Cynthia K. Rigsby, Daniel Kim
NMR IN BIOMEDICINE
(2020)
Article
Radiology, Nuclear Medicine & Medical Imaging
Manuel Schneider, Thomas Benkert, Eddy Solomon, Dominik Nickel, Matthias Fenchel, Berthold Kiefer, Andreas Maier, Hersh Chandarana, Kai Tobias Block
MAGNETIC RESONANCE IN MEDICINE
(2020)
Article
Radiology, Nuclear Medicine & Medical Imaging
C. T. Arendt, K. Eichler, M. G. Mack, D. Leithner, S. Zhang, K. T. Block, Y. Berdan, R. Sader, J. L. Wichmann, T. Gruber-Rouh, T. J. Vogl, M. C. Hoelter
Summary: Dynamic MRI is as reliable as VFS in assessing inadequate velum closure in patients following surgical treatment of VPD, with similar identification of persistent or recurrent VPD cases. MRI showed inferior overall image quality compared to VFS, but still provided accurate anatomical measurements.
EUROPEAN RADIOLOGY
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
T. Demerath, K. Blackham, C. Anastasopoulos, K. T. Block, B. Stieltjes, T. Schubert
MAGNETIC RESONANCE IMAGING
(2020)
Article
Radiology, Nuclear Medicine & Medical Imaging
Eddy Solomon, David S. Rigie, Thomas Vahle, Jan Paska, Jan Bollenbeck, Daniel K. Sodickson, Fernando E. Boada, Kai Tobias Block, Hersh Chandarana
Summary: This study introduces a novel method for detecting respiratory signals using a Pilot-Tone (PT) signal and creating motion-resolved images based on 3D stack-of-stars imaging under free-breathing conditions. The PT signal, generated by a small RF transmitter placed outside the MR bore, is used to reconstruct motion-resolved images from free-breathing scans, leading to improved image quality and clearer anatomical depiction of the lung and liver compared to traditional methods.
MAGNETIC RESONANCE IN MEDICINE
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Li Feng, Fang Liu, Georgios Soultanidis, Chenyu Liu, Thomas Benkert, Kai Tobias Block, Zahi A. Fayad, Yang Yang
Summary: This study developed a rapid free-breathing T1 mapping and fat/water-separated T1 mapping MRI method, accurately evaluated T1 values in different areas, demonstrated good repeatability, and proposed potential applications, providing new possibilities for imaging diagnosis of patients with fatty liver diseases.
MAGNETIC RESONANCE IN MEDICINE
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Hanns C. Breit, Kai T. Block, David J. Winkel, Julian E. Gehweiler, Maurice J. Henkel, Thomas Weikert, Bram Stieltjes, Daniel T. Boll, Tobias J. Heye
Summary: Quantitative assessment of liver fibrosis and cirrhosis using T1 relaxation times showed significant differences in T1 values between healthy liver, acute liver disease, and patients with known fibrosis or cirrhosis. T1 values moderately correlated with the Child-Pugh stage, suggesting T1 mapping is a useful predictor for detecting liver fibrosis and cirrhosis.
EUROPEAN JOURNAL OF RADIOLOGY
(2021)