Conventional Iterative Algorithms Research Articles

Multi-scale perceptual modulation network for low-dose computed tomography denoising.

Low-dose computed tomography (LDCT) reduces radiation exposure, but the introduced noise and artifacts impair its diagnostic accuracy. Convolutional neural networks (CNNs) are widely used for LDCT denoising, but they suffer from a limited receptive field. The use of a larger kernel size can enlarge the receptive field and boost model performance; however, the computational cost of the model greatly increases. We aimed to develop a LDCT denoising CNN with a large receptive field and lower computational complexity. We developed a multi-scale perceptual modulation network (MSPMnet) incorporating a powerful multi-head decomposable convolution (MHDC). To address the high computational complexity of large kernel convolutions, we developed a novel MHDC module that can capture multi-scale features and efficiently expand the receptive field. The MHDC module couples maximum-pooling with three depth-wise convolutions of varying kernel sizes via a channel splitting mechanism, where, unlike conventional CNNs, the two large two-dimensional kernels are each decomposed into a set of cascaded orthogonal one-dimensional kernels to remain lightweight. Further, departing from prior methodologies that apply a uniform kernel size throughout the network, we introduced a receptive field-ramp mechanism that adeptly transitions from local to relatively long-range dependency modeling as the network depth increases, thereby achieving superior performance. The proposed MSPMnet was evaluated on a Mayo Clinic data set with a conventional iterative algorithm, two CNN models, and two Transformer models used for comparison. Compared to the competing baseline methods, the MSPMnet exhibited better performance in both the visual and quantitative assessments. Visually, the MSPMnet preserved the structure, edges, and textures with excellent noise and artifact reduction, generating the denoised images closest to normal-dose computed tomography images. Quantitatively, the MSPMnet had the lowest root mean-square error (RMSE) (8.3094±1.9325) and the highest peak signal-to-noise ratio (PSNR) (33.8525±1.8213 dB), structural similarity index (SSIM) (0.9309±0.0272), and feature similarity index (FSIM) (0.9699±0.0113), demonstrating superior denoising performance. The proposed MSPMnet excelled at LDCT denoising, effectively removing noise and artifacts while preserving edges. Compared to the state-of-the-art CNNs and Transformers, the proposed MSPMnet exhibited superior denoising performance both quantitatively and qualitatively.

Shaping 3D diffraction patterns with a binary aperture

In this Letter, we report an approach for the inverse design of binary apertures to generate desired three-dimensional (3D) diffraction patterns in free space. The approach relies on an optimal accumulation algorithm, aiming to determine the distribution of the binary aperture for 3D target patterns in the regime of Fresnel diffraction. This algorithm features high fidelity for complex inverse design compared with conventional iterative algorithms. To demonstrate the validity of our method, various 2D and 3D patterns are chosen and generated using a digital micromirror device that serves as a reconfigurable binary aperture. Experimentally, the generated diffraction patterns exhibit high fidelity with respect to the target ones, achieving an averaged Pearson correlation coefficient of 0.90 for 2D patterns and 0.87 for 3D patterns, respectively. Our work may find applications in laser beam shaping, structured light illumination, and diffractive optical elements.

Impact on Image Quality and Diagnostic Performance of Dual-Layer Detector Spectral CT for Pulmonary Subsolid Nodules: Comparison With Hybrid and Model-Based Iterative Reconstruction.

To evaluate the image quality and diagnostic performance of pulmonary subsolid nodules on conventional iterative algorithms, virtual monoenergetic images (VMIs), and electron density mapping (EDM) using a dual-layer detector spectral CT (DLSCT). This retrospective study recruited 270 patients who underwent DLSCT scan for lung nodule screening or follow-up. All CT examinations with subsolid nodules (pure ground-glass nodules [GGNs] or part-solid nodules) were reconstructed with hybrid and model-based iterative reconstruction, VMI at 40, 70, 100, and 130 keV levels, and EDM. The CT number, objective image noise, signal-to-noise ratio, contrast-to-noise ratio, diameter, and volume of subsolid nodules were measured for quantitative analysis. The overall image quality, image noise, visualization of nodules, artifact, and sharpness were subjectively rated by 2 thoracic radiologists on a 5-point scale (1 = unacceptable, 5 = excellent) in consensus. The objective image quality measurements, diameter, and volume were compared among the 7 groups with a repeated 1-way analysis of variance. The subjective scores were compared with Kruskal-Wallis test. A total of 198 subsolid nodules, including 179 pure GGNs, and 19 part-solid nodules were identified. Based on the objective analysis, EDM had the highest signal-to-noise ratio (164.71 ± 133.60; P < 0.001) and contrast-to-noise ratio (227.97 ± 161.96; P < 0.001) among all image sets. Furthermore, EDM had a superior mean subjective rating score (4.80 ± 0.42) for visualization of GGNs compared to other reconstructed images (all P < 0.001), although the model-based iterative reconstruction had superior subjective scores of overall image quality. For pure GGNs, the measured diameter and volume did not significantly differ among different reconstructions (both P > 0.05). EDM derived from DLSCT enabled improved image quality and lesion conspicuity for the evaluation of lung subsolid nodules compared to conventional iterative reconstruction algorithms and VMIs.

Withdrawn: Transient modeling of natural gas in pipeline networks by two non-iterative explicit and implicit finite volume methods

This paper describes two non-iterative explicit and implicit finite volume methods (FVM) to calculate the unsteady characteristics of natural gas in distribution systems. The proposed finite volume model could be used to solve the gas network as a whole for each time step. The main advantage of this approach is that it can analyze gas distribution networks with less computational cost. In addition, contrary to the conventional iterative algorithms, it does not consist of any divergence problems (Elman et al., 1996). To achieve this goal, the explicit FVM is derived from the governing equations. Then, the linearization process is performed, and the numerical solution of the linear equations is developed by using the implicit FVM. Three classic test cases are exemplified to check the performance of the proposed methods. The results confirm that both explicit and implicit methods, even in complicated distribution networks, can predict the elements of unsteady natural gas flows. Furthermore, when the governing equations are linearized and the implicit FVM is applied, satisfactory results with less computational cost, compared to the explicit one, are produced.

Enhancing voxel-based dosimetry accuracy with an unsupervised deep learning approach for hybrid medical image registration.

Deformable registration is required to generate a time-integrated activity (TIA) map which is essential for voxel-based dosimetry. The conventional iterative registration algorithm using anatomical images (e.g., computed tomography (CT)) could result in registration errors in functional images (e.g., single photon emission computed tomography (SPECT) or positron emission tomography (PET)). Various deep learning-based registration tools have been proposed, but studies specifically focused on the registration of serial hybrid images were not found. In this study, we introduce CoRX-NET, a novel unsupervised deep learning network designed for deformable registration of hybrid medical images. The CoRX-NET structure is based on the Swin-transformer (ST), allowing for the representation of complex spatial connections in images. Its self-attention mechanism aids in the effective exchange and integration of information across diverse image regions. To augment the amalgamation of SPECT and CT features, cross-stitch layers have been integrated into the network. Two different 177 Lu DOTATATE SPECT/CT datasets were acquired at different medical centers. 22 sets from Seoul National University and 14 sets from Sunway Medical Centre are used for training/internal validation and external validation respectively. The CoRX-NET architecture builds upon the ST, enabling the modeling of intricate spatial relationships within images. To further enhance the fusion of SPECT and CT features, cross-stitch layers have been incorporated within the network. The network takes a pair of SPECT/CT images (e.g., fixed and moving images) and generates a deformed SPECT/CT image. The performance of the network was compared with Elastix and TransMorph using L1 loss and structural similarity index measure (SSIM) of CT, SSIM of normalized SPECT, and local normalized cross correlation (LNCC) of SPECT as metrics. The voxel-wise root mean square errors (RMSE) of TIA were compared among the different methods. The ablation study revealed that cross-stitch layers improved SPECT/CT registration performance. The cross-stitch layers notably enhance SSIM (internal validation: 0.9614vs. 0.9653, external validation: 0.9159vs. 0.9189) and LNCC of normalized SPECT images (internal validation: 0.7512vs. 0.7670, external validation: 0.8027vs. 0.8027). CoRX-NET with the cross-stitch layer achieved superior performance metrics compared to Elastix and TransMorph, except for CT SSIM in the external dataset. When qualitatively analyzed for both internal and external validation cases, CoRX-NET consistently demonstrated superior SPECT registration results. In addition, CoRX-NET accomplished SPECT/CT image registration in less than 6s, whereas Elastix required approximately 50s using the same PC's CPU. When employing CoRX-NET, it was observed that the voxel-wise RMSE values for TIA were approximately 27% lower for the kidney and 33% lower for the tumor, compared to when Elastix was used. This study represents a major advancement in achieving precise SPECT/CT registration using an unsupervised deep learning network. It outperforms conventional methods like Elastix and TransMorph, reducing uncertainties in TIA maps for more accurate dose assessments.

A Neural Network Computational Spectrometer Trained by a Small Dataset with High-Correlation Optical Filters.

A computational spectrometer is a novel form of spectrometer powerful for portable in situ applications. In the encoding part of the computational spectrometer, filters with highly non-correlated properties are requisite for compressed sensing, which poses severe challenges for optical design and fabrication. In the reconstruction part of the computational spectrometer, conventional iterative reconstruction algorithms are featured with limited efficiency and accuracy, which hinders their application for real-time in situ measurements. This study proposes a neural network computational spectrometer trained by a small dataset with high-correlation optical filters. We aim to change the paradigm by which the accuracy of neural network computational spectrometers depends heavily on the amount of training data and the non-correlation property of optical filters. First, we propose a presumption about a distribution law for the common large training dataset, in which a unique widespread distribution law is shown when calculating the spectrum correlation. Based on that, we extract the original dataset according to the distribution probability and form a small training dataset. Then a fully connected neural network architecture is constructed to perform the reconstruction. After that, a group of thin film filters are introduced to work as the encoding layer. Then the neural network is trained by a small dataset under high-correlation filters and applied in simulation. Finally, the experiment is carried out and the result indicates that the neural network enabled by a small training dataset has performed very well with the thin film filters. This study may provide a reference for computational spectrometers based on high-correlation optical filters.

Non-iterative generation of binary amplitude holograms applied to holographic projection with digital micromirror devices

In this work, we implement a fast non-iterative method for the generation of binary amplitude Fresnel holograms and demonstrate their application in a holographic projection scheme based on a digital micromirror device. To achieve this, we perform the binarization of phase-only holograms generated using an optimized Fresnel random phase. We analyze the quality of the resulting binary amplitude holograms and compare them with holograms obtained with the conventional iterative Fresnel algorithm as a function of the number of iterations and the propagation distance. Additionally, we evaluate the diffraction efficiency in both methods. We demonstrate that the holograms produced with our method present significantly advantages in computation speed without a significant reduction in the reconstruction quality. Both experimental and numerical results confirm the effectiveness of our proposal.

Calibration-free quantitative phase imaging in multi-core fiber endoscopes using end-to-end deep learning.

Quantitative phase imaging (QPI) through multi-core fibers (MCFs) has been an emerging in vivo label-free endoscopic imaging modality with minimal invasiveness. However, the computational demands of conventional iterative phase retrieval algorithms have limited their real-time imaging potential. We demonstrate a learning-based MCF phase imaging method that significantly reduced the phase reconstruction time to 5.5 ms, enabling video-rate imaging at 181 fps. Moreover, we introduce an innovative optical system that automatically generated the first, to the best of our knowledge, open-source dataset tailored for MCF phase imaging, comprising 50,176 paired speckles and phase images. Our trained deep neural network (DNN) demonstrates a robust phase reconstruction performance in experiments with a mean fidelity of up to 99.8%. Such an efficient fiber phase imaging approach can broaden the applications of QPI in hard-to-reach areas.

Application of the virtual-fit method for fixed complete denture cases designed on intraoral scans: Effect of cement spacing

ObjectivesTo validate the virtual-fit alignment, analyze the impact of cement spacing on internal/marginal gaps, and correlate results with conventional trueness measures. MethodsFour dental abutment models were scanned using an industrial reference scanner (one time each), Emerald S (three times each), and Medit i700 (three times each) intraoral scanners (IOS). On each IOS scan (n = 24), three complete-arch fixed frameworks were designed with 70 or 140 µm cement space with no marginal space (groups 70 and 140) and 70 µm with an additional 20 µm space, including the margin (group 70+20). Two types of alignment were performed by GOM Inspect software. The reference and IOS scans were aligned through a conventional iterative closest point algorithm (ICP) where the penetration of the two scans was permitted into each other (conventional trueness method). Second, the computer-aided designs were superimposed with the reference scan also using an ICP, but preventing the design from virtual penetration into the model (virtual-fit method). The virtual-fit algorithm was validated by non-penetration alignment of the designs with the IOS scans. Internal and marginal gap was measured between the design and the abutments. The difference between spacing groups was compared by Friedman's test. A statistical correlation (Spearman's Rho Test) was computed between the measured gaps and the conventional trueness method. A significant difference was accepted at p<0.05 after the Bonferroni correction. ResultsThe gaps deviated from the set cement space by 3–13 µm on IOS scans (validation of virtual-fit algorithm). The internal gap of the design on the reference scan was not affected by cement spacing (Emerald S, p = 0.779; Medit i700, p = 0.205). The marginal gap in groups 70 and 70+20 was significantly lower than in group 140 in Emerald S (p<0.05). In Medit i700, it was lower in the 70+20 group than in the group 70 (p<0.01) and in the group 140 (p<0.05). Some Medit i700 scans exhibited high marginal gaps within group 70 but not in groups 70 and 140. The measured gaps correlated significantly (r = 0.51–0.81, p<0.05–0.001) with the conventional trueness but were 2.6–4.6 times higher (p<0.001). ConclusionVirtual-fit alignment can simulate restoration seating. A 20 µm marginal and 90 µm internal spacing could compensate for scan errors up to several hundred micrometers. However, 140 µm internal spacing is counterproductive. The conventional trueness method could only partially predict framework misfit. Clinical significanceThe virtual-fit method can provide clinically interpretable data for intraoral scanners. Emerald S and Medit i700 intraoral scanners are suitable for fabricating complete-arch fixed tooth-supported prostheses. In addition, a slight elevation of spacing at the margin could compensate for moderate inaccuracies in a scan.

Improved artificial gorilla troops optimizer with chaotic adaptive parameters - application to the parameter estimation problem of mixed additive and multiplicative random error models

For mixed additive and multiplicative random error models (MAM models), due to the complex correlation between the parameters and the model power array, derivative operations will be inevitable in the actual calculation. When the observation equation is in nonlinear form, the operations will be more complicated. The swarm intelligence optimization algorithm (SIO) can effectively solve the derivative problem when estimating the nonlinear model parameters using conventional iterative algorithms. However, for different problems, the conventional SIO cannot effectively balance the ability of global and local behavior, resulting in the algorithm falling into prematureness and failing to output effective parameter information. To address the above problems, the improved artificial gorilla troops optimizer (CAGTO) algorithm with chaotic adaptive behavior is proposed. To address the problem that the population generated by the algorithm using pseudo-random numbers in the initialization population phase has poor traversability in the feasible domain, the chaotic sequence is applied to initialize the population instead of pseudo-random number generation to ensure that the population can traverse the feasible domain as much as possible and improve the global search capability of the algorithm. Adaptive parameters that vary linearly and nonlinearly with the algorithm process are constructed to balance the global search and local search ability, while accelerating the convergence speed. Two CAGTO algorithms with different parameter settings are constructed for different problems, and the experimental results show that both CAGTO algorithms can effectively solve the parameter estimation problem of MAM models with different nonlinear forms of observation equations compared with several other comparative algorithms.

Flexible and High-Gain DOFS Deconvolution Based on Data-Driven Denoising Prior

Spatial resolution is an essential parameter for distributed optical fiber sensing (DOFS). Deconvolution is a promising solution to improve spatial resolution because of its good universality and ability to restore infinitely high spatial resolution in theory. However, conventional iterative deconvolution algorithms require manually designed priors, which makes it difficult to restore the signal accurately. Although end-to-end methods based on artificial neural networks provide more accurate results based on data-driven priors, a neural network can only be applied to data with similar characteristics to its training data. For systems with different parameters, the neural network needs to be retrained or fine-tuned, which is time and computational resource consuming. Here, we employ a new deconvolution method for DOFS based on half-quadratic splitting and denoising neural networks that exploits the advantages of both iterative and machine learning methods. Besides, we propose a method to calculate the deconvolution gain to evaluate the deconvolution performance quantitatively. In an experiment with 10 deconvolution ratios, the deconvolution gain of the employed method is 15.8 dB, while that of a conventional iterative method and an end-to-end machine learning method are 7.9 dB and 15 dB, respectively. Moreover, this new deconvolution method can be adapted to data with arbitrary system parameters, making it more flexible than end-to-end neural networks in practical applications.

Evaluation of a Novel Metal Artifact Reduction Algorithm for Reconstruction of Cone Beam CT (CBCT) Images Acquired on a New Imaging System

The presence of metal implants, dental work or prosthetics can produce image artifacts and distort Hounsfield units (HU) in patient CBCT images. The purpose of this work was to characterize a novel metal artifact reduction (MAR) algorithm for reconstruction of CBCT images obtained by a new imaging system and compare improvements in image quality with a conventional iterative reconstruction algorithm. Preliminary analyses show that the novel reconstruction algorithm implemented on the new imaging system produces image quality and HU comparable to standard fan-beam CT MATERIALS/METHODS: A 30 cm (length) x 30 cm (width) x 11 cm (depth) solid water phantom was fitted with a central cavity measuring 5.5 cm x 5.5 cm x 0.5 cm. Metal inserts of copper, aluminum, stainless steel, and titanium (5.5 cm x 5.5 cm x 1 - 5 mm in thickness) and sample dental implant materials (Amalgam, Au, Co-Cr, and Ti 6Al-Yv) were sequentially used to fill the cavity. CBCT images were acquired and reconstructed using the novel MAR and conventional iterative algorithms both on the on the new system. Images were also taken with the cavity filled with solid water for comparison. Eight image quality metrics were measured in six volumes of interest (VOIs) surrounding the cavity and compared to baseline VOI metrics for solid water-only images: HU mean, signal-to-noise ratio, contrast-to-noise ratio, artifact index, standard deviation, range, skewness, and kurtosis. The mean and standard deviation of HU within a shell-structure surrounding the cavity were calculated to evaluate the dependence of these values on distance from the center of the cavity. The artifact volume of the phantom (excluding the cavity) was estimated by calculating the 5th and 95th percentiles of HU in the CBCT of the phantom with the solid water insert, then summing the volume of all voxels with HU values outside this range. The eight image quality metrics extracted from the chosen VOIs were closer to baseline when using the novel MAR compared to conventional iterative reconstruction. It was also determined that mean HU returned to expected solid water background at a shorter distance from metal sample in the novel MAR images, and the standard deviation remained lower for the novel MAR images at all distances from the insert. These experimental outcomes were obtained for all high-density material inserts except for aluminum. The artifact volume was decreased using the novel MAR algorithm for all metal inserts excluding aluminum (p = 0.00013), and all dental implants (p = 0.00028). The novel MAR reconstruction algorithm shows a much lower sensitivity to metal artifacts, and significant improvement in image quality for all metal samples except for aluminum. These results justify the use of the novel MAR reconstruction for patients with metal implants and highlight advantages of the new CBCT imaging system for more accurate imaging and targeted treatment.

Fault estimation and active fault‐tolerant control for a class of nonlinear systems with actuator and sensor faults based on unknown input iterative learning

AbstractIn this paper, fault estimation and active fault‐tolerant control are studied for a class of nonlinear systems with simultaneous actuator and sensor faults, as well as unknown external disturbances. Firstly, the state equation of a class of nonlinear systems is transformed into an augmented system state equation by extending the sensor fault as an auxiliary state. Then, a novel fault estimation observer based on iterative learning with unknown inputs is designed to estimate the system state, as well as actuator and sensor faults. Subsequently, by using the fault estimation information, a dynamic output feedback active fault‐tolerant control scheme is proposed to compensate for the influence of faults on the system. Lyapunov stability theory is used to prove the stability of the closed‐loop system and the convergence of the fault estimation observer. The gain matrices of the fault estimation observer and fault‐tolerant controller are obtained by solving linear matrix inequalities. Furthermore, the paper avoids the use of the norm in the convergence proof of the conventional iterative learning algorithm, which reduces the amount of calculation in the derivation process. Finally, the effectiveness and accuracy of the proposed method are verified through simulation of the DC motor angular velocity system.

Adaptive closed-loop maneuver planning for low-thrust spacecraft using reinforcement learning

Autonomy is an increasingly essential component of future space missions, and new technologies are necessary to accommodate off-nominal occurrences onboard that may pose risks to mission success. Identifying a suitable maneuver plan for onboard, low-thrust mission applications in cislunar space remains challenging, particularly in the case of unanticipated events. This investigation addresses this challenge by demonstrating artificial neural networks as promising tools for accurately estimating startup solutions for a conventional targeting-based iterative guidance and control algorithm. This blended approach produces a robust ‘hybrid’ architecture that simultaneously benefits from the computational simplicity of neural networks and the robustness of differential corrections to satisfy mission requirements. In this paradigm, a multiple shooting scheme is incorporated directly into a reinforcement learning process, tasking the resulting neural network controller with convergence basin identification for trajectory recovery. Rapid low-thrust maneuver planning is demonstrated in a ‘runaway’ spacecraft scenario, where deviation over time from a planned near rectilinear halo orbit path renders stationkeeping ineffective and demands an alternative approach to determine an effective recovery plan. The proposed Neural Network-Initialized Targeting (NNIT) algorithm significantly extends the recovery window compared to a crossing-control orbit maintenance approach and exhibits promising results in identifying both the timing and control components for low-thrust maneuver plans.

Angular Super-Resolution of Multi-Channel APAR in Interference Environments

Aiming to resolve azimuth-dense targets in interference environments, the radar needs to have the ability of single snap echo angular super-resolution with anti-interference. To solve the problem, the angular super-resolution algorithm based on single snap echo while anti-interference with blocking matrix method is studied for active phased array radar (APAR) in this paper. Since the super-resolution ability of the conventional MUSIC algorithm and iterative adaptive algorithm (IAA) algorithm are limited in single snap echo, the iterative re-weighted least squares (IRLS) algorithm under p-norm constraint is proposed. Further, the near main lobe interference suppression ability is enhanced by the adaptive diagonal loading method. The performance differences of IAA and IRLS algorithms for single-target detection and double-target angular super-resolution are analyzed in detail by numerical simulation on three scenes of no interference, side-lobe interference, and near-main-lobe interference. The simulation results show that the proposed algorithm can effectively solve the problem of target angle estimation and super-resolution based on single sample echo in an interference environment, including near main-lobe interference.

Continual Learning-Based Fast Beamforming Adaptation in Downlink MISO Systems

Beamforming is an effective way to improve spectrum efficiency in multi-antenna systems. However, conventional iterative algorithms suffering from high computational delay renders beamforming solutions outdated and deep learning (DL) based realtime solutions suffer from the “task mismatch” problem, leading to severe performance deterioration. To resolve these issues, this letter proposes a beamforming adaptation scheme, named continual learning based beamforming neural network (CL-BNN), which can continuously learn and optimize the downlink beamforming in dynamic environment. The proposed CL-BNN scheme is model-driven and uses domain/expert knowledge to reduce training complexity. By combining our customized selection criteria into the structure of the CL-BNN, which selects the appropriate samples from the existing data and newly received data into a memory set for model training, taking into account the objective function, such that the tradeoff between maintaining performance of the trained task and adapting to the new task is achieved. Simulation results show that the proposed CL-BNN scheme achieves good adaptability in dynamic environments with low complexity.

Fast generation of arbitrary optical focus array

We report a novel method to generate arbitrary optical focus arrays (OFAs). Our approach rapidly produces computer-generated holograms (CGHs) to precisely control the positions and the intensities of the foci. This is achieved by replacing the fast Fourier transform (FFT) operation in the conventional iterative Fourier-transform algorithm (IFTA) with a linear algebra one, identifying/removing zero elements from the matrices, and employing a generalized weighting strategy. On the premise of accelerating the calculation speed by >70 times, we demonstrate OFA with 99% intensity precision in the experiment. Our method proves effective and is applicable to systems in which real-time OFA generation is essential.

TransUNet-based inversion method for ghost imaging

Ghost imaging (GI), which employs speckle patterns and bucket signals to reconstruct target images, can be regarded as a typical inverse problem. Iterative algorithms are commonly considered to solve the inverse problem in GI. However, high computational complexity and difficult hyperparameter selection are the bottlenecks. An improved inversion method for GI based on the neural network architecture TransUNet is proposed in this work, called TransUNet-GI. The main idea of this work is to utilize a neural network to avoid issues caused by conventional iterative algorithms in GI. The inversion process is unrolled and implemented on the framework of TransUNet. The demonstrations in simulation and physical experiment show that TransUNet-GI has more promising performance than other methods.

Deep learning image reconstruction algorithm reduces image noise while alters radiomics features in dual-energy CT in comparison with conventional iterative reconstruction algorithms: a phantom study.

To compare image quality between a deep learning image reconstruction (DLIR) algorithm and conventional iterative reconstruction (IR) algorithms in dual-energy CT (DECT) and to assess the impact of these algorithms on radiomics robustness. A phantom with clinical-relevant densities was imaged on seven DECT scanners with the same voxel size using typical abdominal-pelvis examination protocols. On one DECT scanner, raw data were reconstructed using both conventional IR (adaptive statistical iterative reconstruction-V, ASIR-V) and DLIR. Nine sets of corresponding images were generated on other six DECT scanners using scanner-equipped conventional IR. Regions of interest were delineated through rigid registrations. Image quality was compared. Pyradiomics platform was used for radiomics feature extraction. Test-retest repeatability was assessed by Bland-Altman analysis for repeated scans. Inter-reconstruction algorithm reproducibility between conventional IR and DLIR was tested by intraclass correlation coefficient (ICC) and concordance correlation coefficient (CCC). Inter-scanner reproducibility was evaluated by coefficient of variation (CV) and quartile coefficient of dispersion (QCD). Robust features were identified. DLIR significantly improved image quality. Ninety-four radiomics features were extracted and nine features were considered as robust. 93.87% features were repeatable between repeated scans. ASIR-V images showed higher reproducibility to other conventional IR than DLIR (ICC mean, 0.603 vs 0.558, p = 0.001; CCC mean, 0.554 vs 0.510, p = 0.004). 7.45% and 26.83% features were reproducible among scanners evaluated by CV and QCD, respectively. DLIR improves quality of DECT images but may alter radiomics features compared to conventional IR. Nine robust DECT radiomics features were identified. • DLIR improves DECT image quality in terms of signal-to-noise ratio and contrast-to-noise ratio compared with ASIR-V and showed the highest noise reduction rate and lowest peak frequency shift. • Most of radiomics features are repeatable between repeated DECT scans, while inter-reconstruction algorithm reproducibility between conventional IR and DLIR, and inter-scanner reproducibility, are low. • Although DLIR may alter radiomics features compared to IR algorithms, nine radiomics features survived repeatability and reproducibility analysis among DECT scanners and reconstruction algorithms, which allows further validation and clinical-relevant analysis.

Flow field tomography with uncertainty quantification using a Bayesian physics-informed neural network

We report a new approach to flow field tomography that uses the Navier–Stokes and advection–diffusion equations to regularize reconstructions. Tomography is increasingly employed to infer 2D or 3D fluid flow and combustion structures from a series of line-of-sight (LoS) integrated measurements using a wide array of imaging modalities. The high-dimensional flow field is reconstructed from low-dimensional measurements by inverting a projection model that comprises path integrals along each LoS through the region of interest. Regularization techniques are needed to obtain realistic estimates, but current methods rely on truncating an iterative solution or adding a penalty term that is incompatible with the flow physics to varying degrees. Physics-informed neural networks (PINNs) are new tools for inverse analysis that enable regularization of the flow field estimates using the governing physics. We demonstrate how a PINN can be leveraged to reconstruct a 2D flow field from sparse LoS-integrated measurements with no knowledge of the boundary conditions by incorporating the measurement model into the loss function used to train the network. The resulting reconstructions are remarkably superior to reconstructions produced by state-of-the-art algorithms, even when a PINN is used for post-processing. However, as with conventional iterative algorithms, our approach is susceptible to semi-convergence when there is a high level of noise. We address this issue through the use of a Bayesian PINN, which facilitates comprehensive uncertainty quantification of the reconstructions, enables the use of a more intuitive loss function, and reveals the source of semi-convergence.