Plant 3D reconstruction using optical imaging often suffers from incomplete point clouds due to viewpoint occlusion and sensor limitations. This incompleteness hinders accurate structural representation and subsequent feature extraction for plant analysis. To address these challenges, we propose a Multi-Scale Dual-Discriminator Generative Adversarial Network (MSDDG) for plant point cloud completion. A multi-scale point cloud generator (MSPG) that integrates local and global features from raw incomplete point clouds is used for MSDDG to reconstruct complete shapes. The dual-discriminators-a multi-view projected silhouette discriminator and a spatial distance discriminator-are designed to ensure geometric realism and spatial plausibility from multiple perspectives. To train MSDDG, we created the Plant4L dataset containing four plant species (sunflower, pumpkin, luffa, and eggplant) with high-quality 3D models augmented via 3D thin plate spline transformations and virtual occlusion simulation to generate incomplete point clouds and multi-view silhouettes. Experimental results on Plant4L demonstrate that MSDDG achieves superior completion performance, with Chamfer Distance (CD), Hausdorff Distance (HD), and Uniformity Chamfer Distance (UCD) all below 0.41. Comparative evaluations confirm MSDDG's superiority over previous point cloud completion methods. The application of MSDDG for 3D reconstruction from single view further validate its effectiveness in restoring occluded plant structures.

Murray-law based quantitative flow ratio (μFR) enables rapid fractional flow reserve (FFR) computation from invasive coronary angiography (ICA) using a single projection, but the influence of plaque eccentricity on its diagnostic accuracy remains unclear. To investigate whether eccentric plaque impacts the diagnostic accuracy of single-view μFR. We performed a blinded analysis of the prospective CAREER trial database, enrolling patients with 30%-90% diameter stenosis on coronary computed tomography angiography (CCTA) who underwent μFR and FFR assessments within 30 days. ICA were acquired using standardized, protocol-specified projections. CCTA images were analyzed using dedicated software and co-registered with ICA. For each μFR-identified lesion, plaque eccentricity index (PEI) and lumen asymmetry index (LAI) were computed across all cross-sections and averaged to yield PEI and LAI per-vessel. Vessels were classified as having eccentric/concentric plaques using median PEI, and subclassified as having asymmetric/symmetric lumens using median LAI. Among 231 vessels (201 patients), median μFR and FFR were 0.84 and 0.83, respectively. PEI and LAI moderately correlated (ρ = 0.46, p < 0.001). Limits of agreement between μFR and FFR were wider in eccentric versus concentric plaques (standard deviation 0.08 vs. 0.06; p = 0.003), mainly driven by presence of asymmetric lumens (standard deviation 0.09 vs. 0.06 in symmetric lumens; p = 0.029). μFR had comparable AUC for predicting FFR ≤ 0.80 between concentric plaques and eccentric plaques with symmetric lumens (0.94 vs. 0.95; p = 0.909). The diagnostic accuracy of single-view μFR, derived from standardized angiographic projections, was moderately affected by eccentric plaques, with the effect primarily attributable to asymmetric lumens.

A novel computational angiographic microcirculatory resistance (AMR) derived from a single angiographic view presents a feasible alternative to the pressure wire-based index of microcirculatory resistance. However, its prognostic significance in patients undergoing percutaneous coronary intervention (PCI) remains insufficiently established. This is a post hoc analysis of 3404 patients undergoing PCI from the FAVOR III China (Comparison of Quantitative Flow Ratio Guided and Angiography Guided Percutaneous Intervention in Patients With Coronary Artery Disease) trial. Pre- and post-PCI AMR were measured in target vessels, with percentage change in AMR before and after PCI calculated as (100×[post-PCI AMR-pre-PCI AMR]/pre-PCI AMR). The primary model used was the log-rank test, and the proportional hazards model was also used to assess the association between AMR and the 3-year risk of major adverse cardiac events, defined as a composite of all-cause death, myocardial infarction, or ischemia-driven revascularization. Patients with percentage change in AMR before and after PCI ≥85 (23.7%) versus <85 (76.3%) had comparable baseline characteristics but received more and longer stents per patient. Overall major adverse cardiac events risk was similar between groups (14.8% versus 12.4%; hazard ratio [HR], 1.18 [0.95-1.45]; log-rank P=0.064). However, in patients with post-PCI AMR ≥250, percentage change in AMR before and after PCI ≥85 showed a significant increase in the major adverse cardiac events risk (16.3% versus 10.8%; HR, 1.52 [1.14-2.04]), contrasting with no difference when post-PCI AMR <250 (12.3% versus 13.4%; HR, 0.89 [0.63-1.25]; Pinteraction=0.019). In patients undergoing PCI from the FAVOR III China population, significant AMR elevation (percentage change in AMR before and after PCI ≥85) in target vessels alone did not predict outcomes, but in the subgroup with post-PCI AMR ≥250 it identified patients at increased 3-year cardiovascular risk. https://www.clinicaltrials.gov; Unique identifier: NCT03656848.

Abstract In this review article, an exhaustive study has been conducted regarding several scientific papers on diverse 3D viewing systems and 3D projection techniques developed over the time span of 2020 to 2025. With advancements in computer vision, computer graphics, and artificial intelligence, there has been a significant utilization of 3D visualization and projection techniques in multiple industries including virtual reality, augmented reality, medicine, robots, entertainment, and intelligent space construction. In this literature review, 30 research papers are mentioned based on different themes of 3D reconstruction, Gaussian splatting, neural rendering, depth perception, panoramas, structured light setup, and real-time rendering techniques. The new trends in the fields of single view reconstruction, Multiview reconstruction, sparse view reconstruction, dynamic reconstruction, and material-aware rendering are explored in these papers. However, although all these advancements have been made, there are still issues of computation efficiency, memory efficiency, robustness, and ethics that must be addressed. Contemporary 3D viewing and projection techniques have established themselves as efficient tools but there is always room for improvement. Keywords 3D Visualization, 3D Reconstruction, Gaussian Splatting, Virtual Reality, Augmented Reality, Depth Estimation, View Synthesis

Abstract Reconstructing fast-ion velocity distributions from collective Thomson scattering (CTS) spectra is an ill-posed inverse problem due to the spectral nonlinearity, strong parameter coupling and measurement noise. To mitigate the ill-posed problems in single-view inversions, a hybrid spectrum–parameter conditioned encoder (HSPCE) is proposed to reconstruct two-dimensional fast-ion velocity distributions based on the electromagnetic forward model. For feasibility validation, one-dimensional inversion is firstly carried out using an electrostatic CTS model under HL-3 operating parameters. Compared with the Least Squares with Nuisance Parameters (LSN), the machine-learning approach demonstrates markedly improved robustness and accuracy, maintaining an R2 of 0.950 with 10% Gaussian noise. Extending to two dimensions, the velocity distribution is represented as an image and reduced in dimensionality through principal component analysis (PCA), with additional soft constraints applied in both coefficient and pixel spaces to preserve physical consistency. Benchmarking against the 1D baseline model, HSPCE shows that the higher structural similarity (SSIM) and lower normalized root mean square error (NRMSE) with different gaussian noise, with 0.930 and 0.043 at noise level of 0.1 respectively. Further analysis indicates that the fraction of fast ions plays an important role in enabling the network to extract reliable fast-ion information, with higher fractions yielding clearer reconstructions and reduced uncertainty. Overall, the proposed framework suggests that neural networks offer a promising and robust approach for improving the interpretability and reliability of fast-ion diagnostics based on CTS in magnetically confined plasmas.

Standardized 2D photography plays an essential role in dermatologic practice, supporting longitudinal documentation, patient monitoring, and consensus-based clinical scoring. However, photographs taken from limited views often suffer from reduced anatomical context and missing body-location information. 3D representations enable unified spatial interpretation of multi-view imagery. Recent developments in computer vision have made it feasible to infer dense correspondences between 2D images and a 3D human mesh. In this study, we explored integrating 2D dermatological images with a 3D surface model using DensePose, a deep learning-based human dense correspondence framework. This creates an anatomically grounded representation that supports mesh-level analyses and recovers spatial context for each image. We use a dataset including four full body photographs (front, back, and each side) from each of 147 subjects with chronic graft-versus-host disease, for a total of 588 images. Our method integrates these multiple 2D full-body photographs captured across varied body shapes and camera angles into a 3D mesh. We further showed that the resulting 3D mesh enables quantification of the extent to which individual 2D images, or their combinations, represent the complete body surface. On average, a single full body view captures 28% of the body surface, while adding a second, third, and fourth view increases average coverage to 50%, 72%, and 80%, respectively. To assess spatial consistency, we annotated up to 10 anatomical landmarks per patient on 80 images across 20 patients and reported a median pairwise geodesic distance between corresponding landmarks of 4.6 cm. These findings can guide how dermatology images are captured and support future opportunities in monitoring, education, and communication using existing infrastructure.

Tricuspid regurgitation (TR) is often incidentally detected at intraoperative transesophageal echocardiography (TEE), resulting in possible changes in the surgical plan. The aim of this study was to compare 2D and 3D measurements of TR vena contracta width (VCW) and the degree of TR severity using TEE from the three standard mid-esophageal (ME) views. In a prospective observational study, we analyzed 3D and 2D TEE datasets from 30 adult patients undergoing elective tricuspid valve (TV) repair. 2D and 3D TEE color flow Doppler (CFD) loops of the TV in the three standard ME views (4Chamber, RV inflow-outflow (inflow) and modified bicaval) were recorded immediately after induction of anesthesia. VCW from single views and the average of the measurement of biplane VCW (2D biplane VCW) from each standard view were compared with the maximum and minimum diameters of the 3D vena contracta area (VCA) and their average (3D average VCW). TR severity classification was compared between 2D biplane VCW and 3D average VCW and VCA. Correlation between measurements was analyzed using Pearson coefficient and agreement assessed using the Bland-Altman method. Cohen's Kappa correlation was used to assess TR severity concordance. Biplane VCW in all three ME views underestimated 3D average VCW measurement, with VCW from ME inflow view showing the best agreement. VCW measurements in single standard views systematically underestimated the maximum 3D VCA diameter. We detected very good agreement in TR grading between 3D average VCW and 3D VCA, and an underestimation by 2D biplane VCW (moderate agreement for inflow and fair for the other views). Intra- and inter-observer correlation when repeating 2D measurements was more reliable than that for 3D measurements. Our study shows that 2D biplane VCW from the ME inflow view best agrees with 3D average VCW and allows the most accurate classification of TR severity compared to 3D average VCW.

The ambiguity in paragraph 2 of Section 38 of the Electronic Commerce Law regarding the liability of e-commerce service providers for non-compliance has led to persistent disputes in academic and practical circles. This paper explores what constitutes a more reasonable corresponding liability for platform operators. Theoretically, two major approaches exist: one advocates for a multi-faceted liability framework by comparing the platform with its internal operational conduct, while the other proposes a single, clearly defined liability. It is submitted that the non-compliance of the online trading platform with its review obligations cannot be regarded as active conduct for the purposes of joint and several liability. Rather, it can be considered passive inaction. Conversely, the single liability view seeks to define the liability type from multiple perspectives. In judicial practice, due to the difficulty of proving platform operators' fault in reality, courts generally tend to mitigate their liability. Overall, after analyzing differing perspectives in theory and practice, the author argues that interpreting platform operators' “bearing corresponding liability” as supplementary liability is more reasonable.

Physiological assessment of left circumflex coronary artery. A bicentric validation and outcome study using μFR.

Recent advancements in traffic forecasting leverage spatiotemporal features through recurrent and graph neural networks, showing promising results. However, these approaches typically require extensive manual hyperparameter tuning, rely on a single data view, and often overlook the dynamic nature of traffic patterns, which change at every timestamp. Analyzing spatial interactions among traffic nodes is critical for improving accuracy, yet adaptive graph convolutional networks that respond to these dynamic interactions remain underutilized. Recent advances in bio-inspired algorithms, such as particle swarm optimization (PSO), have demonstrated their power in optimizing and automating model generation and parameter tuning processes. These algorithms can systematically explore a wide range of model configurations to identify optimal strategies, thereby improving learning and data view identification. Specifically, bio-inspired algorithms enhance the adaptive capabilities of traffic forecasting models by dynamically adjusting to changing traffic patterns and sensor data. To address these challenges, this study presents two novel bio-inspired traffic forecasting models designed to improve forecasting accuracy and facilitate more efficient traffic management. The proposed framework presents two primary model variants: (1) PSO-optimized Long Short-Term Memory (LSTM) and Bidirectional LSTM (Bi-LSTM) models for temporal forecasting, and (2) an integrated PSO-optimized GCN-LSTM-BiLSTM model that incorporates both spatiotemporal features and historical data for comprehensive prediction. This framework highlights the flexibility of our evolutionary optimization methodology while addressing the limitation of single-view analysis, which fails to capture the evolving nature of traffic patterns, leading to suboptimal predictions. The present approach automates hyperparameter selection and dynamically weights temporal, bidirectional, and spatial views, enabling robust adaptation to real-time traffic fluctuations. Extensive experiments on PeMSD07 and PeMSD08 datasets demonstrate the framework’s effectiveness: the optimized LSTM and Bi-LSTM models achieved accuracies of 99.21% and 99.15%, respectively, while the integrated spatiotemporal model attained a mean average error of 0.1948, outperforming existing baselines. The results confirm that PSO-driven automation and adaptive multi-view analysis significantly improve prediction accuracy and robustness, offering a scalable solution for intelligent traffic management systems.

Accurate assessment of the functional significance of coronary artery stenoses is essential for guiding revascularization decisions and improving clinical outcomes in patients with coronary artery disease (CAD). While invasive wire-based fractional flow reserve (FFR) remains the gold standard for physiological lesion assessment, its adoption in routine clinical practice is limited by procedural complexity, patient discomfort, time consumption, and cost. These limitations have driven the development of angiography-derived FFR techniques that enable physiological evaluation without pressure wires or pharmacologic hyperaemia. Recent advances in computational modelling, artificial intelligence, and image processing have facilitated the estimation of FFR from conventional coronary angiography, including approaches that require only a single angiographic view. Single-view angiography-derived FFR methods-such as Murray law-based quantitative flow ratio (µQFR), FFR2D, Angio-iFR/FFR, sAccuFFR, and X1-FFR-aim to simplify workflow while maintaining diagnostic accuracy. Among these, µQFR has demonstrated the most consistent validation against invasive FFR across a broad range of clinical scenarios, including complex lesions, severe aortic stenosis, multivessel disease, and acute coronary syndromes. This review summarizes the principles, validation data, clinical applications, and limitations of single-view angiography-derived FFR technologies and highlights their potential to expand the adoption of physiology-guided coronary intervention.

Fiber bundle imaging systems suffer from sampling artifacts such as honeycomb patterns due to their discrete and non-uniform fiber layout, fundamentally limiting image resolution. Conventional reconstruction methods rely on precise calibration of the fiber layout or learning from paired datasets, both of which have limited generalization across imaging setups and require sample-specific preparation. We present an unsupervised method for reconstructing high-resolution images using a burst of misaligned frames that does not require known fiber layout, paired training data, or per-sample calibration. Our approach jointly solves motion estimation and image reconstruction through test-time training. We model each burst frame as a deformed observation of a single canonical view, parameterizing the underlying motion with a coordinate-based network. A second coordinate-based network learns a joint super-resolved scene representation shared across aligned frames. Both networks are trained jointly end-to-end without paired ground truth or external supervision. Simulation and experimental results demonstrate that our method robustly removes fiber bundle artifacts and generalizes to various sample types. We also released a benchmark dataset for optical fiber bundle imaging to facilitate future research.

ABSTRACT Engineering is considered important in solving unsustainability. However, it is a complex problem that must be viewed, analysed and studied from various perspectives and taking with the contribution of various areas of knowledge. This work studied the use of interdisciplinarity as a contribution to interconnect ethics and sustainability with technical-scientific contents of electrical engineering. The research intended to use interdisciplinarity to help engineering students recognise that engineering is not ethically neutral, and that, therefore, the problems (and solutions) must also be analysed from an ethical and sustainability perspective. A framework was developed, and a pedagogical activity using interdisciplinarity was applied. Results show that, after the activity, students recognise that ethical values influence calculations in the area of electrical installations; and move from a single view to identify different alternatives, perspectives, motivations and multiple objectives. This leads to studying more alternatives and hopefully better overall technical solutions.

Using 2D video analysis and model based image matching to measure joint angles for forensic biomechanical analysis.

Molecular foundation models hold promise to provide accurate predictions for a large and diverse set of downstream tasks in bio-medical research. Quality molecular representations are key and foundation model development has typically focused on a single representation or molecular view, which may have strengths or weaknesses on a given task. We develop Multi-view Molecular Embedding with Late Fusion (MMELON), an approach that integrates pre-trained graph, image and text foundation models and may be readily extended to additional views and models. The multi-view model performs robustly and is validated on over 120 tasks, including molecular solubility, ADME properties, and activity against G Protein-Coupled receptors (GPCRs). The GPCR model array is leveraged to perform a virtual screen in search of ligands binding to Alzheimer's disease related GPCRs. We identify a number of such targets and employ the multi-view model to select strong binders from a compound screen. Predictions are validated through structure-based modeling and identification of key bindingmotifs.

Protein structures are crucial in understanding the function, mechanism and disease-causing variants of proteins within any living cell. A number of experimental techniques are employed by researchers to determine such structures. Through structure inspection in molecular viewers, combined with supporting biochemical and biophysical experiments, scientists are able to identify the function of a protein, its reaction mechanism and effects caused by sequence variation. These detailed findings, supported by experimental results, are documented by being described in the scientific literature and by making the accompanying data open source. However, it has become increasingly difficult for a reader, in particular a non-expert, to access the correct additional information and assess the validity of the conclusions drawn based on experimental results. A reader is often required to resort to a number of different software packages to access the different data types. Here, we present a first-of-its-kind implementation of an artificial intelligence- and text-mining-supported software tool that allows the association of mentions in the text of one or more specific protein residues with their corresponding counterparts in the respective protein structure or structures. Our application allows a researcher to explore a residue of interest in the context of a publication and its respective protein structure, supported by its experimental evidence, in a single view. We describe model implementation, annotation extraction, downstream processing, dissemination and visualization at the IUCr and PDBe. The application presented is primarily aimed at readers of IUCr publications and users visiting the PDBe entry pages. However, we believe that in the future our application will be a valuable tool for reviewers of new submissions to IUCr journals and may even be useful as a curation tool involving the authors of a publication as annotation validators.

The study aimed to compare the radiation dose between digital mammography (DM) and digital breast tomosynthesis (DBT) and to correlate it with compressed breast thickness (CBT) and breast density. Patient age, CBT, and breast density, kVp, mAs, average glandular dose (AGD), and entrance skin dose (ESD), for 701 cases were retrieved. The radiation dose difference between DM in craniocaudal (CC) and DBT in mediolateral oblique (MLO) views was determined across different CBT and density categories. Multiple linear regression analysis was performed to identify the AGD and ESD predictors. The overall AGD of MLO DBT (2.46 mGy) was higher than that of CC DM (2.30 mGy), P&lt;0.001. Similarly, the ESD of MLO DBT (7.67 mGy) was higher than that of CC DM (6.32 mGy), P&lt;0.001. CBT was the primary determinant of AGD in both DM and DBT, with standardized coefficients beta of 0.4 and 0.6, respectively. CBT also was the primary determinant of ESD in both DM and DBT with standardized coefficients beta of 0.2 and 0.9, respectively. The current study found that the AGD of the MLO view of DBT was higher than that of the CC view of DM. However, the overall AGD was still within the ACR-recommended AGD for single view.

Chronic wound classification remains a major challenge in clinical settings due to the subjective nature of visual assessment, inconsistent documentation practices, and the presence of overlapping visual features across wound types. These limitations contribute to delayed or inaccurate diagnosis, which in turn compromises treatment outcomes. Despite the growing use of deep learning in medical imaging, most existing models in wound analysis rely on single-view input, limiting their ability to generalise under real-world clinical conditions. In this study, we present a deep learning-based framework that leverages input-level diversity through ensemble and multi-branch convolutional neural networks (CNNs). Our approach uses three image variants: original, Lab colour space contrast limited adaptive histogram equalisation (CLAHE), and maximum green greyscale channel CLAHE —processed through identical pre-trained CNNs. We conducted experiments on a combined wound image dataset drawn from the publicly available advancing the Zenith of Healthcare dataset and an independent proprietary dataset, as well as external validation using the Medetec wound image database. Our Input-diverse ensemble framework, based on weighted probability fusion achieved an accuracy of 89.23% and F 1-score of 82.95%, outperforming individual model variants and demonstrating strong generalisation during external validation. Comparative analysis also shows that our model meets and exceeds the performance of existing state-of-the-art methods reported on similar classification tasks. The proposed method introduces a lightweight yet effective solution for improving classification consistency across multiple wound types without requiring complex multimodal data or additional clinical metadata. Its design offers a practical step forward for integrating deep learning into clinical workflows, enabling scalable, interpretable, and more reliable wound assessment tools for frontline healthcare professionals.

Reconstructing hand and object shapes from a single view during interaction remains challenging due to severe mutual occlusion and the need for high physical plausibility. To address this, we propose a novel framework for hand–object interaction reconstruction based on holistic, multi-stage collaborative optimization. Unlike methods that process hands and objects independently or apply constraints as late-stage post-processing, our model progressively enforces physical consistency and geometric accuracy throughout the entire reconstruction pipeline. Our network takes an RGB-D image as input. An adaptive feature fusion module first combines color and depth information to improve robustness against sensing uncertainties. We then introduce structural priors for 2D pose estimation and leverage texture cues to refine depth-based 3D pose initialization. Central to our approach is the iterative application of a dense mutual attention mechanism during sparse-to-dense mesh recovery, which dynamically captures interaction dependencies while refining geometry. Finally, we use a Signed Distance Function (SDF) representation explicitly designed for contact surfaces to prevent interpenetration and ensure physically plausible results. Through comprehensive experiments, our method demonstrates significant improvements on the challenging ObMan and DexYCB benchmarks, outperforming state-of-the-art techniques. Specifically, on the ObMan dataset, our approach achieves hand CDh and object CDo metrics of 0.077 cm2 and 0.483 cm2, respectively. Similarly, on the DexYCB dataset, it attains hand CDh and object CDo values of 0.251 cm2 and 1.127 cm2, respectively.

In Inflammatory Language, Una Stojnić and Ernie Lepore argue that no extant theory of slurs can explain slurs’ hyperprojectivity, emphasizing their difficulties in accounting for acoustic and phonological resemblance cases in which a word merely sounds like a slur. Further, all content theories confront the Specificity Problem, the charge that the content view’s content, whatever it is, is too specific to encompass the full range of competent weapon uses of slurs. One half of this paper concerns hyperprojectivity. I argue that there is a gap in Inflammatory Language’s overarching dialectic that results from excluding a range of theories. Some theories of slurs are what I call single mechanism views: they aim to explain all the phenomena with a single explanatory mechanism. Multiple mechanism views exploit more than one. Within Inflammatory Language, multiple mechanism theories are bypassed. Yet multiple mechanism theories possess resources to explain slurs’ hyperprojectivity. The other half of this paper addresses the Specificity Problem. I argue that a view I have developed in previous writings, Identity Expressivism, does not succumb to the problem. I craft a version of the Specificity Problem tailor-made for the theory and rooted in Stojnić and Lepore case against other expressivist theories. Identity Expressivism is, I argue, uncompromised by the Specificity Problem.