Article
Engineering, Electrical & Electronic
Anish Lahiri, Gabriel Maliakal, Marc L. Klasky, Jeffrey A. Fessler, Saiprasad Ravishankar
Summary: This work focuses on image reconstruction when both the number of available CT projections and the training data is extremely limited. A sequential reconstruction approach is adopted, using an adversarially trained shallow network for 'destreaking' followed by a data-consistency update in each stage. To address the challenge of limited data, image subvolumes are used for training and patch aggregation during testing. To handle the computational challenge of 3D reconstruction, a hybrid 3D-to-2D mapping network is used for the 'destreaking' part. Comparisons with other methods indicate the potential of the proposed method in scenarios with highly limited projections and training data.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2023)
Article
Computer Science, Interdisciplinary Applications
Sushanth G. Sathyanarayana, Zhuoying Wang, Naidi Sun, Bo Ning, Song Hu, John A. Hossack
Summary: This study develops a sparse modeling method for blood flow quantification, which can accurately recover substantially downsampled data and shows good performance in both in vitro and in vivo experiments.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Computer Science, Artificial Intelligence
Jingke Zhang, Qiong He, Yang Xiao, Hairong Zheng, Congzhi Wang, Jianwen Luo
Summary: A new method for PWUS image reconstruction from RF data is proposed using a DNN and a self-supervised learning scheme, significantly reducing computational time. Validation with simulation data shows excellent performance in terms of reconstruction time, spatial resolution, and CNR.
MEDICAL IMAGE ANALYSIS
(2021)
Article
Engineering, Electrical & Electronic
Mengjie Guo, Hengrong Lan, Changchun Yang, Jiang Liu, Fei Gao
Summary: This paper presents a novel signal processing method to improve sparse sensor data in photoacoustic (PA) imaging and proposes an Attention Steered Network (AS-Net) algorithm for PA image reconstruction with multi-feature fusion. Experimental results show that the method achieves superior reconstructions at a faster speed.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2022)
Article
Computer Science, Information Systems
Xinxin Zuo, Sen Wang, Jiangbin Zheng, Weiwei Yu, Minglun Gong, Ruigang Yang, Li Cheng
Summary: This paper introduces a novel approach to reconstructing 3D human body shapes from a sparse set of RGBD frames using a single RGBD camera. The proposed framework effectively addresses the challenge of fusing these sparse frames into a canonical 3D model, demonstrating superior performance through empirical evaluations on synthetic and real datasets. Additionally, the flexibility of the framework suggests potential applications beyond shape reconstruction, such as reshaping and reposing to a new avatar.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Biology
Liyue Shen, Wei Zhao, Dante Capaldi, John Pauly, Lei Xing
Summary: This study establishes a geometry-informed deep learning framework for ultra-sparse 3D tomographic image reconstruction. The seamless inclusion of geometric priors is shown to be essential for enhancing imaging performance and provides new avenues for data-driven biomedical imaging.
COMPUTERS IN BIOLOGY AND MEDICINE
(2022)
Article
Computer Science, Information Systems
Yongbo Wang, Gaofeng Chen, Tao Xi, Zhaoying Bian, Dong Zeng, Habib Zaidi, Ji He, Jianhua Ma
Summary: Sparse-view helical CT images suffer from noise, artifacts, and severe anatomical distortions, reducing the applicability of existing reconstruction algorithms. A novel TDATV model is proposed for SHCT reconstruction, utilizing tensor decomposition and anisotropic total variation regularization to reduce HCT radiation dose. Results show the potential of SHCT in achieving ultra-low dose CT examinations.
Article
Engineering, Electrical & Electronic
Simon Grosche, Andy Regensky, Alexander Sinn, Jurgen Seiler, Andre Kaup
Summary: Non-regular three-quarter sampling using L-shaped pixels has been shown to improve image sensor quality, and a faster version of the reconstruction algorithm, RL-JSDE, provides significant speedups on both CPU and GPU without sacrificing image quality.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS
(2022)
Article
Optics
Graeme E. Johnstone, Johannes Herrnsdorf, Martin D. Dawson, Michael J. Strain
Summary: Challenging imaging applications that require ultra-short exposure times or imaging in photon-starved environments often have extremely low numbers of photons per pixel (<1 photon per pixel). To improve the image quality in such photon-sparse images, post-processing techniques, such as Bayesian retrodiction and bilateral filtering, can be used to estimate the number of photons detected and improve the spatial distributions in single-photon imaging applications. In this study, we demonstrate that at high frame rates (>1 MHz) and low incident photon flux (<1 photon per pixel), image post-processing techniques can provide better grayscale information and spatial fidelity of reconstructed images compared to simple frame averaging, with up to a 3-fold improvement in SSIM.
Article
Computer Science, Interdisciplinary Applications
Wenjun Xia, Zexin Lu, Yongqiang Huang, Yan Liu, Hu Chen, Jiliu Zhou, Yi Zhang
Summary: This paper introduces a parameter-dependent framework that trains a reconstruction network with data from multiple alternative geometries and dose levels simultaneously, reducing extra training costs for multiple geometries and dose levels.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2021)
Article
Computer Science, Artificial Intelligence
Pablo Rodrigo Gantier Cadena, Yeqiang Qian, Chunxiang Wang, Ming Yang
Summary: Event-based cameras offer advantages over traditional cameras, but utilizing the data they produce is challenging due to the unique nature of event sensors. Neural networks have led to significant advances in event-based image reconstruction, with the new SPA DE-E2VID model showing improved video quality. The model also features faster training time and allows reconstruction without a temporal loss function, demonstrating promising results for event camera technology.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Optics
Qile Zhao, Xu Ma, Gonzalo R. Arce, Zhiqiang Wang
Summary: A model-driven deep learning approach is proposed in this study to significantly improve the computational efficiency and accuracy of compressive X-ray tomosynthesis reconstruction. By jointly optimizing coding masks and neural network parameters, the degrees of optimization freedom are effectively increased, leading to a significant improvement in computational efficiency and performance of reconstruction results.
Article
Computer Science, Artificial Intelligence
Wenqi Huang, Ziwen Ke, Zhuo-Xu Cui, Jing Cheng, Zhilang Qiu, Sen Jia, Leslie Ying, Yanjie Zhu, Dong Liang
Summary: In this study, a model-based low-rank plus sparse network (L+S-Net) is proposed for dynamic MR image reconstruction, achieving clear separation of the L component and S component through learned soft singular value thresholding and demonstrating global convergence. Experimental results show that the proposed model outperforms the current state-of-the-art methods on retrospective and prospective cardiac cine datasets, with great potential for high acceleration factors.
MEDICAL IMAGE ANALYSIS
(2021)
Article
Geochemistry & Geophysics
Hang Wang, Yunfeng Chen, Yapo Abole Serge Innocent Oboue, Ray Abma, Zhicheng Geng, Sergey Fomel, Yangkang Chen
Summary: This paper proposes a simple and effective framework for fast reconstruction and denoising of undersampled 5-D seismic data. It tackles the limitations of data acquisition geometry and complex geological structures. The proposed framework has a low computational cost and preserves signal fidelity.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Information Systems
Rui Shi, Honglong Zheng, Xianguo Tuo, Changming Wang, Jianbo Yang, Yi Cheng, Mingzhe Liu, Songbai Zhang
Summary: Tomographic Gamma Scanning (TGS) is a crucial non-destructive technique for analyzing radioactive waste drums. By applying the TVM method, the MLEM-TVM and ART-TVM reconstruction methods show improved accuracy and signal-to-noise ratio compared to traditional algorithms, with MLEM-TVM achieving the best results in image quality. The TVM method not only enhances the TGS image resolution but also saves scanning time through sparse projection sampling.
Article
Optics
Steven Johnson, Alex McMillan, Stefan Frick, John Rarity, Miles Padgett
Summary: A limitation of free-space optical communications is the ease of interception, which can be overcome by hiding information within background optical noise. We demonstrate image transfer over free-space using a photon-pair source emitting two correlated beams. One beam contains image information with added noise, while the other correlated beam serves as a heralding trigger to differentiate the image signal from background noise. The system utilizes spontaneous parametric down-conversion and a gated intensified camera to extract the image from the noise.
Article
Multidisciplinary Sciences
Stirling Scholes, German Mora-Martin, Feng Zhu, Istvan Gyongy, Phil Soan, Jonathan Leach
Summary: Single-Photon Avalanche Detector (SPAD) arrays are cutting-edge technology that can rapidly generate depth images with millimeter precision. They are crucial for autonomous systems as they provide guidance and situational awareness. This study establishes a numerical procedure to determine the depth imaging limits of SPAD arrays under real-world conditions, allowing for accurate and cost-effective performance evaluation without field testing. This procedure has applications in object detection, tracking, and imaging systems for various scenarios.
SCIENTIFIC REPORTS
(2023)
Article
Optics
Stirling Scholes, Lehloa Mohapi, Jonathan Leach, Andrew Forbes, Angela Dudley
Summary: In this study, a Liquid Crystal on Silicon Spatial Light Modulator (LCoS-SLM) is used to imitate the mechanical design of a deformable mirror, and the effect of the number of mirror segments and their geometrical structure on resulting structured modes is quantitatively analysed. This approach can serve as a test bed prior to designing a deformable mirror for high power beam shaping.
APPLIED PHYSICS B-LASERS AND OPTICS
(2023)
Article
Optics
Mohamed Amir Alaa Belmekki, Jonathan Leach, Rachael Tobin, Gerald S. Buller, Stephen McLaughlin, Abderrahim Halimi
Summary: 3D single-photon LiDAR imaging is crucial for many applications, but requires analysis of low signal-to-noise ratio returns and high data volume. This paper proposes a multiscale approach for 3D surface detection, reducing data volume while obtaining background-free surfaces for depth and reflectivity inference. A hierarchical Bayesian model is also introduced for 3D reconstruction and spectral classification. Results demonstrate the superiority of these approaches compared to state-of-the-art algorithms.
Article
Physics, Multidisciplinary
Suraj Goel, Max Tyler, Feng Zhu, Saroch Leedumrongwatthanakun, Mehul Malik, Jonathan Leach
Summary: Efficient manipulation, sorting, and measurement of optical modes and single-photon states are achieved in this study. The researchers use a specially designed multiplane light converter to simultaneously and efficiently sort nonorthogonal, overlapping states of light encoded in the transverse spatial degree of freedom. This has implications for optimal image identification and classification in optical networks.
PHYSICAL REVIEW LETTERS
(2023)
Article
Chemistry, Analytical
Gyles E. Cozier, Rachael C. Andrews, Anca Frinculescu, Ranjeet Kumar, Benedict May, Tom Tooth, Peter Collins, Andrew Costello, Tom S. F. Haines, Tom P. Freeman, Ian S. Blagbrough, Jennifer Scott, Trevor Shine, Oliver B. Sutcliffe, Stephen M. Husbands, Jonathan Leach, Richard W. Bowman, Christopher R. Pudney
Summary: Synthetic cannabinoids (SCs) are a class of new psychoactive substances (NPS) primarily used in prisons and homeless communities in the U.K. SCs have severe side effects and are associated with numerous deaths. The detection of SCs is challenging due to their chemical diversity and their adsorption onto physical matrices. This study presents a portable and rapid generic test for SCs using fluorescence analysis.
ANALYTICAL CHEMISTRY
(2023)
Article
Optics
Konstantin Y. Bliokh, Ebrahim Karimi, Miles J. Padgett, Miguel A. Alonso, Mark R. Dennis, Angela Dudley, Andrew Forbes, Sina Zahedpour, Scott W. Hancock, Howard M. Milchberg, Stefan Rotter, Franco Nori, Sahin K. Ozdemir, Nicholas Bender, Hui Cao, Paul B. Corkum, Carlos Hernandez-Garcia, Haoran Ren, Yuri Kivshar, Mario G. Silveirinha, Nader Engheta, Arno Rauschenbeutel, Philipp Schneeweiss, Juergen Volz, Daniel Leykam, Daria A. Smirnova, Kexiu Rong, Bo Wang, Erez Hasman, Michela F. Picardi, Anatoly Zayats, Francisco J. Rodriguez-Fortuno, Chenwen Yang, Jie Ren, Alexander B. Khanikaev, Andrea Alu, Etienne Brasselet, Michael Shats, Jo Verbeeck, Peter Schattschneider, Dusan Sarenac, David G. Cory, Dmitry A. Pushin, Michael Birk, Alexey Gorlach, Ido Kaminer, Filippo Cardano, Lorenzo Marrucci, Mario Krenn, Florian Marquardt
Summary: Structured waves are found in all areas of wave physics, both classical and quantum, where the wavefields are inhomogeneous and cannot be approximated by a single plane wave. These complex wavefields with inhomogeneities are crucial in various fields such as nanooptics, photonics, quantum matter waves, acoustics, water waves, etc. This Roadmap surveys the role of structured waves in wave physics, providing background, current research, and anticipating future developments.
Review
Optics
Robert h. Hadfield, Jonathan Leach, Fiona Fleming, Douglas j. Paul, Chee hing Tan, Jo shien Ng, Robert k. Henderson, Gerald s. Buller
Summary: The development of single-photon detectors with picosecond timing resolution has driven progress in time-correlated single-photon counting applications, including quantum optics, life sciences, and remote sensing. Advanced optoelectronic device architectures offer high-performance single-pixel devices and the ability to scale up to detector arrays, increasing single-photon sensitivity.
Article
Optics
German Mora-Martin, Stirling Scholes, Lice Ruget, Robert Henderson, Jonathan Leach, Istvan Gyongy
Summary: In this paper, a 3D convolutional neural network (CNN) is trained using synthetic depth sequences to denoise and upscale (x4) depth data. Experimental results using synthetic and real ToF data demonstrate the effectiveness of the approach. With GPU acceleration, the approach achieves processing speeds of >30 frames per second, making it suitable for low-latency imaging.
Article
Quantum Science & Technology
Chane Moodley, Alice Ruget, Jonathan Leach, Andrew Forbes
Summary: In order to improve image acquisition speed, researchers implemented four machine learning algorithms and trained them on a dataset with noise and blur. The results showed that logistic regression algorithm achieved a 10x speed up in image acquisition time with a prediction accuracy of 99%. This method does not require image denoising or enhancement prior to recognition, reducing training and implementation time, as well as computational intensity, making it suitable for real-time quantum imaging and recognition of light sensitive structures.
ADVANCED QUANTUM TECHNOLOGIES
(2023)
Article
Optics
Imogen Morland, Hanna Ostapenko, Feng Zhu, Derryck T. Reid, Jonathan Leach
Summary: In this study, correlated photon pairs were generated at 790nm wavelength using a compact GHz-rate cavity laser. The indistinguishability of the photons produced by SPDC was verified using Hong-Ou-Mandel interferometry, and a dip in coincidence counts with a visibility of 81.8% was observed.
Article
Optics
Sara Restuccia, Graham M. Gibson, Leroy Cronin, Miles J. Padgett
Summary: This study demonstrates the measurement of optical activity in a sample using an unpolarized light source, with the help of a polarization-entangled photon source. This approach allows for low light measurement and the analysis of samples that may be perturbed by polarized light.
Proceedings Paper
Optics
Stirling Scholes, Alice Ruget, German M. Martin, Feng Zhu, Istvan Gyongy, Robert K. Henderson, Jonathan Leach
Summary: The development of single-photon avalanche diode (SPADs) arrays in time-of-flight imaging systems has enabled the application of 3D imaging for drone identification, orientation, and segmentation. By combining the imaging capability of SPAD sensors with the classification capabilities of convolutional neural networks, an accurate determination of drone pose in flight can be achieved, with prediction accuracy of over 90% after training.
ADVANCED PHOTON COUNTING TECHNIQUES XVI
(2022)
Article
Optics
Steven Johnson, Alex McMillan, Yril Torre, Stefan Frick, John Rarity, Miles Padgett
Summary: Traditional remote sensing applications based on pulsed laser illumination are not suitable for covert operation. We present a method that uses correlated photon-pairs to perform single-pixel imaging, suppressing background light effect and improving signal-to-noise ratio.
Article
Computer Science, Information Systems
Stirling Scholes, Alice Ruget, German Mora-Martin, Feng Zhu, Istvan Gyongy, Jonathan Leach
Summary: This paper presents a method for fully characterizing drones in flight using a CNN with a decision tree and ensemble structure. The system can determine the type and orientation of drones, as well as perform segmentation for classification of different body parts. The researchers also provide a computer model for generating large quantities of accurately labeled photo-realistic training data, which allows the system to accurately characterize real drones in flight.