Fast Reconstruction Research Articles

An optimized EBRSA-Bi LSTM model for highly undersampled rapid CT image reconstruction

COVID-19 has spread all over the world, causing serious panic around the globe. Chest computed tomography (CT) images are integral in confirming COVID positive patients. Several investigations were conducted to improve or maintain the image reconstruction quality for the sample image reconstruction. Deep learning (DL) methods have recently been proposed to achieve fast reconstruction, but many have focused on a single domain, such as the image domain of k-space. In this research, the highly under-sampled enhanced battle royale self-attention based bi-directional long short-term (EBRSA-bi LSTM) CT image reconstruction model is proposed to reconstruct the image from the under-sampled data. The research is adapted with two phases, namely, pre-processing and reconstruction. The extended cascaded filter (ECF) is proposed for image pre-processing and tends to suppress the noise and enhance the reconstruction accuracy. In the reconstruction model, the battle royale optimization (BrO) is intended to diminish the loss function of the reconstruction network model and weight updation. The proposed model is tested with two datasets, COVID-CT- and SARS-CoV-2 CT. The reconstruction accuracy of the proposed model with two datasets is 93.5 % and 97.7 %, respectively. Also, the image quality assessment parameters such as Peak-Signal to Noise Ratio (PSNR), Root Mean Square Error (RMSE) and Structural Similarity Index metric (SSIM) are evaluated, and it yields an outcome of (45 and 46 dB), (0.0026 and 0.0022) and (0.992, 0.996) with two datasets.

Deep learning-based fusion of widefield diffuse optical tomography and micro-CT structural priors for accurate 3D reconstructions.

Widefield illumination and detection strategies leveraging structured light have enabled fast and robust probing of tissue properties over large surface areas and volumes. However, when applied to diffuse optical tomography (DOT) applications, they still require a time-consuming and expert-centric solving of an ill-posed inverse problem. Deep learning (DL) models have been recently proposed to facilitate this challenging step. Herein, we expand on a previously reported deep neural network (DNN) -based architecture (modified AUTOMAP - ModAM) for accurate and fast reconstructions of the absorption coefficient in 3D DOT based on a structured light illumination and detection scheme. Furthermore, we evaluate the improved performances when incorporating a micro-CT structural prior in the DNN-based workflow, named Z-AUTOMAP. This Z-AUTOMAP significantly improves the widefield imaging process's spatial resolution, especially in the transverse direction. The reported DL-based strategies are validated both in silico and in experimental phantom studies using spectral micro-CT priors. Overall, this is the first successful demonstration of micro-CT and DOT fusion using deep learning, greatly enhancing the prospect of rapid data-integration strategies, often demanded in challenging pre-clinical scenarios.

Compression Reconstruction Network with Coordinated Self-Attention and Adaptive Gaussian Filtering Module

Although compressed sensing theory has many advantages in image reconstruction, its reconstruction and sampling time is very long. Fast reconstruction of high-quality images at low measurement rates is the direction of the effort. Compressed sensing based on deep learning provides an effective solution for this. In this study, we propose an attention-based compression reconstruction mechanism (ACRM). The coordinated self-attention module (CSAM) is designed to be embedded in the main network consisting of convolutional blocks and utilizes the global space and channels to focus on key information and ignore irrelevant information. An adaptive Gaussian filter is proposed to solve the loss of multi-frequency components caused by global average pooling in the CSAM, effectively supplementing the network with different frequency information at different measurement rates. Finally, inspired by the basic idea of the attention mechanism, an improved loss function with attention mechanism (AMLoss) is proposed. Extensive experiments show that the ACRM outperforms most compression reconstruction algorithms at low measurement rates.

Deep compressed sensing MRI via a gradient-enhanced fusion model.

Compressed sensing has been employed to accelerate magnetic resonance imaging by sampling fewer measurements. However, conventional iterative optimization-based CS-MRI are time-consuming for iterative calculations and often share poor generalization ability on multicontrast datasets. Most deep-learning-based CS-MRI focus on learning an end-to-end mapping while ignoring some prior knowledge existed in MR images. We propose an iterative fusion model to integrate the image and gradient-based priors into reconstruction via convolutional neural network models while maintaining high quality and preserving the detailed information aswell. We propose a gradient-enhanced fusion network (GFN) for fast and accurate MRI reconstruction, in which dense blocks with dilated convolution and dense residual learnings are used to capture abundant features with fewer parameters. Meanwhile, decomposed gradient maps containing obvious structural information are introduced into the network to enhance the reconstructed images. Besides this, gradient-based priors along directions X and Y are exploited to learn adaptive tight frames for reconstructing the desired image contrast and edges by respective gradient fusion networks. After that, both image and gradient priors are fused in the proposed optimization model, in which we employ the l2 -norm to promote the sparsity of gradient priors. The proposed fusion model can effectively help to capture edge structures in the gradient images and to preserve more detailed information of MRimages. Experimental results demonstrate that the proposed method outperforms several CS-MRI methods in terms of peak signal-to-noise (PSNR), the structural similarity index (SSIM), and visualizations on three sampling masks with different rates. Noteworthy, to evaluate the generalization ability, the proposed model conducts cross-center training and testing experiments for all three modalities and shares more stable performance compared than other approaches. In addition, the proposed fusion model is applied to other comparable deep learning methods. The quantitative results show that the reconstruction results of these methods are obviouslyimproved. The gradient-based priors reconstructed from GFNs can effectively enhance the edges and details of under-sampled data. The proposed fusion model integrates image and gradient priors using l2 -norm can effectively improve the generalization ability on multicontrast datasetsreconstruction.

Deep learning particle identification in LHCb RICH

The use of Ring Imaging Cherenkov detectors (RICH) offers a powerful technique for identifying the particle species in particle physics. These detectors produce 2D images formed by rings of individual photons superimposed on a background of photon rings from other particles.The RICH particle identification (PID) is essential to the LHCb experiment at CERN. While the current PID algorithm has performed well during LHC data-taking periods between 2010 to 2018, its complexity poses a challenge for LHCb computing infrastructure upgrades towards multi-core architectures. The high particle multiplicity environment of future LHC runs strongly motivates shifting towards high-throughput computing for the online event reconstruction.In this contribution, we introduce a convolutional neural network (CNN) approach to particle identification in LHCb RICH. The CNN takes binary input images from the two RICH detectors to classify particle species. The input images are polar-transformed sub-sections of the RICH photon-detection planes. The model is hyperparameter-optimised and trained on classification accuracy with simulated collision data for the upcoming LHC operation starting in 2022. The PID performance of the CNN is comparable to the conventional algorithm, and its simplicity renders it suitable for fast online reconstruction through parallel processing.We show that under conditions of reduced combinatorial background, as expected from the introduction of timing resolution to the RICH detectors in future upgrades, the network achieves a particle identification performance close to 100 %, with simultaneous misclassification of the most prevalent particle species approaching 0 %.

What Binarization Method Is the Best for Amplitude Inline Fresnel Holograms Synthesized for Divergent Beams Using the Direct Search with Random Trajectory Technique?

Fast reconstruction of holographic and diffractive optical elements (DOE) can be implemented by binary digital micromirror devices (DMD). Since micromirrors of the DMD have two positions, the synthesized DOEs must be binary. This work studies the possibility of improving the method of synthesis of amplitude binary inline Fresnel holograms in divergent beams. The method consists of the modified Gerchberg-Saxton algorithm, Otsu binarization and direct search with random trajectory technique. To achieve a better quality of reconstruction, various binarization methods were compared. We performed numerical and optical experiments using the DMD. Holograms of halftone image with size up to 1024 × 1024 pixels were synthesized. It was determined that local and several global threshold methods provide the best quality. Compared to the Otsu binarization used in the original method of the synthesis, the reconstruction quality (MSE and SSIM values) is improved by 46% and the diffraction efficiency is increased by 27%.

Lensless holographic microscope with a time and memory-saving algorithm for large-volume imaging of organoids.

Organoids, the 3D culture systems derived from stem cells, are promising models for human organs. However, organoid study requires large-volume imaging with single cell resolution, which is beyond the spatial bandwidth limit of conventional optical microscopy. Herein, we propose a lensless holographic microscope empowered with a time and memory-saving algorithm. It solves the trade-off between the imaging field of view, resolution, and processing speed, and provides a practical tool for the study of organoids. We first build a compact microscopy system using a multi-angle LED illumination scheme and an on-chip structure. Then we develop a fast angular spectrum formula for fast reconstruction of oblique-illuminated coaxial holography under the under-sampling condition. Additionally, we derive a multi-angle illuminated filtered backpropagation algorithm to achieve high-precision and slice-wise recovery of 3D structures of objects. The reconstruction process demands only 1/50 of the memory required by a traditional optical diffraction tomography algorithm. Experimental results indicate that the proposed method can achieve 6.28 mm × 4.71 mm × 0.37 mm volume imaging within 104 s. Through the standardized polystyrene beads test, we demonstrate that the proposed microscope has micrometer-scale resolution in both lateral and axial directions. In addition, the 3D imaging results of salivary gland organoids show great application prospects of the proposed method in the field of living biological sampling imaging.

Fast Quantum State Reconstruction via Accelerated Non-Convex Programming

We propose a new quantum state reconstruction method that combines ideas from compressed sensing, non-convex optimization, and acceleration methods. The algorithm, called Momentum-Inspired Factored Gradient Descent (MiFGD), extends the applicability of quantum tomography for larger systems. Despite being a non-convex method, MiFGD converges provably close to the true density matrix at an accelerated linear rate asymptotically in the absence of experimental and statistical noise, under common assumptions. With this manuscript, we present the method, prove its convergence property and provide the Frobenius norm bound guarantees with respect to the true density matrix. From a practical point of view, we benchmark the algorithm performance with respect to other existing methods, in both synthetic and real (noisy) experiments, performed on the IBM’s quantum processing unit. We find that the proposed algorithm performs orders of magnitude faster than the state-of-the-art approaches, with similar or better accuracy. In both synthetic and real experiments, we observed accurate and robust reconstruction, despite the presence of experimental and statistical noise in the tomographic data. Finally, we provide a ready-to-use code for state tomography of multi-qubit systems.

Full-field x-ray fluorescence imaging using a Fresnel zone plate coded aperture

We present an approach to full-field x-ray fluorescence imaging that uses a Fresnel zone plate as a coded aperture positioned between the object and detector. For this particular coded aperture, the incoherent image formation is identical to the coherent propagation image (inline hologram) of an equivalent absorption object. In the first experiment, we show that prior knowledge of the coded aperture structure and position is no longer required. Fast and robust reconstruction algorithms well known from coherent imaging can be applied. We demonstrate the approach using x-ray fluorescence excited by a collimated Cu- K α beam at an ordinary laboratory x-ray source. Images are recorded with a field of view of 1 m m 2 and resolution of 35 µm. The simple and efficient imaging scheme is well suited for further dissemination of x-ray fluorescence imaging and translation to a broader user community, which could take advantage of full-field imaging with chemical contrast. More generally, the scheme also enables full-field imaging of other incoherent radiation processes for which lenses are either not available at all or not with sufficient numerical apertures. Examples include Compton imaging, imaging of inelastic neutron scattering, or γ -ray imaging in nuclear medicine.

On-chip micro-ring resonator array spectrum detection system based on convex optimization algorithm

Abstract We propose an all on-chip micro-ring resonator array spectrum detection system (MRRAS). Micro-ring resonator array as the core is used to construct the transmission matrix of the system. The theoretical analysis of the spectrum detection system is completed with waveguide transmission theory and spectrum construction method based on convex optimization algorithm. In the experiment, we obtain the priori information of the transmission matrix of the system, then detect the output intensity of unknown spectrum through MRRAS, and construct the under-determined matrix equations when the number of micro-rings is much smaller than that of reconstructed wavelengths. Convex optimization algorithm is employed to obtain the least norm solution of the under-determined matrix equations, which enables fast spectrum reconstruction. The experimental results show that the spectrum detection system is constructed using three micro-ring resonators with 4 μm radius, enabling the compact footprint. In addition, the silicon nitride based photonic platform is fully compatible with standard complementary metal oxide semiconductor (CMOS) processes. The system operating bandwidth is more than 12 nm and the resolution is better than 0.17 nm.

Alternative deep learning method for fast spatial-frequency shift imaging microscopy.

Spatial-frequency shift (SFS) imaging microscopy can break the diffraction limit of fluorescently labeled and label-free samples by transferring the high spatial-frequency information into the passband of microscope. However, the resolution improvement is at the cost of decreasing temporal resolution since dozens of raw SFS images are needed to expand the frequency spectrum. Although some deep learning methods have been proposed to solve this problem, no neural network that is compatible to both labeled and label-free SFS imaging has been proposed. Here, we propose the joint spatial-Fourier channel attention network (JSFCAN), which learns the general connection between the spatial domain and Fourier frequency domain from complex samples. We demonstrate that JSFCAN can achieve a resolution similar to the traditional algorithm using nearly 1/4 raw images and increase the reconstruction speed by two orders of magnitude. Subsequently, we prove that JSFCAN can be applied to both fluorescently labeled and label-free samples without architecture changes. We also demonstrate that compared with the typical spatial domain optimization network U-net, JSFCAN is more robust to deal with deep-SFS images and noisy images. The proposed JSFCAN provides an alternative route for fast SFS imaging reconstruction, enabling future applications for real-time living cell research.

Deep supervised dictionary learning by algorithm unrolling-Application to fast 2D dynamic MR image reconstruction.

Unrolled neural networks (NNs) have been extensively applied to different image reconstruction problems across all imaging modalities. A key component of the latter is that they allow for physics-informed learning of the regularization method, which is parametrized by the NN. However, due to the lack of understanding of deep NNs from a theoretical point of view, unrolled NNs are still black-boxes when the regularizers are given by deep NNs, for example, U-Nets. Dictionarylearning (DL) is a well-established regularization method, which is based on learning a transform to sparsely approximate the signals of interest. Typically, DL-based image reconstruction either employs a dictionary, which was pretrained on a set of patches which were extracted from ground-truth images or a dictionary which is jointly trained during the reconstruction. However, in both cases, the used DL-algorithms are not designed to take into account the reconstruction problem or the underlying physical model, which describes the imaging process. In this work, we propose a DL-algorithm based on unrolled NNs to overcome these limitations. We construct an unrolled NN, which corresponds to an unrolled DL-based reconstruction algorithm and train the unrolled NN to optimize its weights, that is, the atoms of the dictionary, by back-propagation in a supervised manner. Further, we propose a new way to employ a 2D dictionary in the spatio-temporal domain. We tested and evaluated the method on an accelerated cardiac cine MR image reconstruction problem using 216/36/36 dynamic images for training, validation, and testing and compared it to two well-known state-of-the-art approaches for cardiac cine MRI based on deep iterative CNNs. Further, we analyze the obtained dictionaries in terms of dictionary-coherence and structure of the atoms. Last, we compare the reported methods in terms of stability by applying them to an entirely different dataset consisting of 49 different testimages. The investigated physics-informed DL-approach yields significantly more accurate reconstructions compared to the DL-method, which uses dictionaries obtained by decoupled pretraining, thereby providing an improvement of up to 4.90 dB in terms of PSNR and 5% in terms of SSIM. Further, the proposed spatio-temporal 2D dictionary outperforms the 1D and 3D dictionaries by preventing smoothing of image details while still accurately removing undersampling artifacts and noise resulting in an increase of up to 1.10 dB in terms of PSNR and 4% in terms of SSIM. Although being surpassed by the CNNs on the first dataset, the proposed NNs-based DL method is more stable compared to the latter approach and yields comparable results on the second dataset. Last, it has the advantage of being entirely interpretable in eachcomponent. The presented physics-informed NN can be used as training algorithm for a classical and interpretable data-driven regularization method based on a learned dictionary, which can then not only be linked to the considered data but also to the reconstruction method that the NNdefines.

Fast iterative reconstruction for photoacoustic tomography using learned physical model: Theoretical validation

Iterative reconstruction has demonstrated superior performance in medical imaging under compressed, sparse, and limited-view sensing scenarios. However, iterative reconstruction algorithms are slow to converge and rely heavily on hand-crafted parameters to achieve good performance. Many iterations are usually required to reconstruct a high-quality image, which is computationally expensive due to repeated evaluations of the physical model. While learned iterative reconstruction approaches such as model-based learning (MBLr) can reduce the number of iterations through convolutional neural networks, it still requires repeated evaluations of the physical models at each iteration. Therefore, the goal of this study is to develop a Fast Iterative Reconstruction (FIRe) algorithm that incorporates a learned physical model into the learned iterative reconstruction scheme to further reduce the reconstruction time while maintaining robust reconstruction performance. We also propose an efficient training scheme for FIRe, which releases the enormous memory footprint required by learned iterative reconstruction methods through the concept of recursive training. The results of our proposed method demonstrate comparable reconstruction performance to learned iterative reconstruction methods with a 9x reduction in computation time and a 620x reduction in computation time compared to variational reconstruction.

Optimization of communication topology for persistent formation in case of communication faults

To address the optimization problem of communication topology for persistent formation in the case of communication faults such as link interruption, transmitter failure, and receiver failure a two-stage model including fast reconstruction of communication topology and re-optimization of communication topology is constructed. Then, a fast reconstruction algorithm of communication topology for persistent formation (FRA-CT-PF), based on optimally rigid graph, arc addition operation, and path reversal operation, is proposed, which can quickly generate a feasible reconstructed communication topology after communication faults, thus ensuring the safety of the agents and maintaining the formation shape of persistent formation. Furthermore, a re-optimization algorithm of communication topology for persistent formation (ROA-CT-PF), based on agent position exchange, is proposed, which can further obtain a reoptimized communication topology to minimize the formation communication cost while still maintaining the formation shape of persistent formation. The time complexities of these two algorithms are also analyzed. Finally, the effectiveness of the above algorithms is verified by numerical experiments. Compared with existing algorithms, FRA-CT-PF can always obtain feasible reconstructed communication topology in much less time under all communication fault scenarios, and ROA-CT-PF can obtain a reoptimized communication topology to further reduce the formation communication cost in a shorter time.

Reconstruction of pointwise sources in a time-fractional diffusion equation

This paper is concerned with an inverse pointwise source problem for the time-fractional diffusion equation in the two-dimensional case. The source term to be identified models the action of a finite number of small particles. Each particle is assumed to be no larger than a single point, characterized by its location and intensity. Both theoretical and numerical aspects are discussed. In the theoretical part, we analyse the well-posedness of the Dirac time-fractional diffusion problem. For the inverse problem, we establish that the unknown point sources can be uniquely identified from local measured data and we derive a local Lipschitz stability result. In the numerical part, we develop a fast and accurate reconstruction approach. The unknown pointwise sources are characterized as solution to an optimization problem minimizing a tracking-type functional. A noniterative reconstruction algorithm is devised, allowing us to determine the number, locations and intensities of the pointwise sources. The efficiency of the proposed approach is confirmed by some numerical examples.

In Situ Coupling of Carbon Dots with Co-ZIF Nanoarrays Enabling Highly Efficient Oxygen Evolution Electrocatalysis.

Metal-organic frameworks (MOFs) are regarded as one promising class of precatalysts for electrocatalytic oxygen evolution reaction (OER), yet most of them suffer from poor conductivity and lack of coordinatively unsaturated metal sites, which hinders the fast electrochemical reconstruction and thus a poor OER activity. To address this issue, a unique heterocomposite has been constructed by in situ inserting carbon dots (CDs) into cobalt-based zeolitic imidazolate framework (Co-ZIF) nanosheet arrays (Co-ZIF/CDs/CC) in the presence of carbon cloth (CC) via one-pot coprecipitation for alkaline OER. Benefiting from the synergism between CDs and Co-ZIF subunits such as superior conductivity, strong charge interaction as well as abundant metal sites exposure, the Co-ZIF/CDs/CC exhibits an enhanced promotion effect for OER and contributes to the deep phase transformation from CDs-coupled Co-ZIF to CDs-coupled active CoOOH. As expected, the achieved Co-ZIF/CDs/CC only requires an overpotential of 226mV to deliver 10mA cm-2 in 1.0M KOH, which is lower than that of Co-ZIF/CC and superior to most previously reported CC-supported MOF precatalysts. Moreover, it can also maintain a large current density of 100mA cm-2 for 24h with negligible activity decay.

A Correlation Coefficient Sparsity Adaptive Matching Pursuit Algorithm

This paper presents a correlation coefficient sparsity adaptive matching pursuit (CCSAMP) algorithm for practical compressed sensing (CS). The sparsity adaptive matching pursuit (SAMP) has been enhanced using the CCSAMP algorithm. The CCSAMP's capacity to accurately reconstruct the signal with fewer repetitions is its most novel characteristic when compared to other state-of-the-art SAMP enhancement methods. This makes it a candidate for many practical applications that need fast reconstruction. The proposed algorithm constructs two correlation vectors, which represent the input signals recovered from the support set and candidate set. The step size is transformed by their Pearson correlation coefficients (PCCS). Compared to the residual energy, the correlation coefficient is more sensitive. The CCSAMP reduces the number of iterations while maintaining the SAMP's capability of signal reconstruction without prior knowledge of the sparsity. Simulation shows that the CCSAMP can significantly reduce the number of iterations compared to the SAMP algorithm. The CCSAMP can be used for radar detection, radar 3D imaging, and other fields where fast and accurate reconstruction of signals is required.

3D Visualization of Pulmonary Vessel Based on Low-Cost Segmentation and Fast Reconstruction

Real-time visual-aided navigation and path strategy for pneumonoconiosis and efficient 3D visualization of pulmonary vessels are of great research and clinical significance in the treatment of lung diseases. The complex structure of lung tissue limits the application of deep learning in pulmonary vascular visualization due to the lack of vascular labeling datasets. Also, the existing methods have large computational complexity and are low efficiency. This study proposes a method for high-quality 3D visualization of pulmonary vessels based on low-cost segmentation and fast reconstruction, consisting of three steps: 1) Pulmonary vessel feature extraction from lung CT images using self-supervised learning, 2) Segmentation of pulmonary sparse vessels in lung CT images using self-supervised transfer learning, and 3) 3D reconstruction of pulmonary vessels based on segmentation results of step (2) using interpolation. The accuracy of pulmonary vascular contour segmentation was improved from 91.31% using the sparse coding to 98.65% using our proposed method (27,270 test sample points); the classifier evaluation accuracy was improved from 95.33% to 98.26%, and the average running time of the model with the test set data was 44 ms per slice. the segmentation results can automatically generate a complete vascular tree model with an average time of 10.8s ± 1 1.6s. The results demonstrate that the proposed method provides fast and accurate 3D visualization of pulmonary vessels, and is promising for more precise and reliable information for pneumonoconiosis patients.

Fast Mesh Reconstruction from Single View Based on GCN and Topology Modification

3D reconstruction based on single view aims to reconstruct the entire 3D shape of an object from one perspective. When existing methods reconstruct the mesh surface of complex objects, the surface details are difficult to predict and the reconstruction visual effect is poor because the mesh representation is not easily integrated into the deep learning framework; the 3D topology is easily limited by predefined templates and inflexible, and unnecessary mesh self-intersections and connections will be generated when reconstructing complex topology, thus destroying the surface details; the training of the reconstruction network is limited by the large amount of information attached to the mesh vertices, and the training time of the reconstructed network is too long. In this paper, we propose a method for fast mesh reconstruction from single view based on Graph Convolutional Network (GCN) and topology modification. We use GCN to ensure the generation of high-quality mesh surfaces and use topology modification to improve the flexibility of the topology. Meanwhile, a feature fusion method is proposed to make full use of the features of each stage of the image hierarchically. We use 3D open dataset ShapeNet to train our network and add a new weight parameter to speed up the training process. Extensive experiments demonstrate that our method can not only reconstruct object meshes on complex topological surfaces, but also has better qualitative and quantitative results.

Fast Reconstruction for Deep Learning PET Head Motion Correction.

Head motion correction is an essential component of brain PET imaging, in which even motion of small magnitude can greatly degrade image quality and introduce artifacts. Building upon previous work, we propose a new head motion correction framework taking fast reconstructions as input. The main characteristics of the proposed method are: (i) the adoption of a high-resolution short-frame fast reconstruction workflow; (ii) the development of a novel encoder for PET data representation extraction; and (iii) the implementation of data augmentation techniques. Ablation studies are conducted to assess the individual contributions of each of these design choices. Furthermore, multi-subject studies are conducted on an 18F-FPEB dataset, and the method performance is qualitatively and quantitatively evaluated by MOLAR reconstruction study and corresponding brain Region of Interest (ROI) Standard Uptake Values (SUV) evaluation. Additionally, we also compared our method with a conventional intensity-based registration method. Our results demonstrate that the proposed method outperforms other methods on all subjects, and can accurately estimate motion for subjects out of the training set. All code is publicly available on GitHub: https://github.com/OnofreyLab/dl-hmc_fast_recon_miccai2023.