3D Field Research Articles

Multifunctional Metasurface Tuning by Liquid Crystals in Three Dimensions.

Optical metasurfaces present remarkable opportunities for manipulating wave propagation in unconventional ways, surpassing the capabilities of traditional optical devices. In this work, we introduce and demonstrate a multifunctional dynamic tuning of dielectric metasurfaces containing liquid crystals (LCs) through an effective three-dimensional (3D) control of the molecular orientation. We theoretically and experimentally study the spectral tuning of the electric and magnetic resonances of dielectric metasurfaces, which was enabled by rotating an external magnetic field in 3D. Our approach allows for the independent control of the electric and magnetic resonances of a metasurface, enabling multifunctional operation. The magnetic field tuning approach eliminates the need for the pre-alignment of LCs and is not limited by a finite set of directions in which the LC molecules can be oriented. Our results open new pathways for realizing dynamically reconfigurable metadevices and observing novel physical effects without the usual limitations imposed by the boundary conditions of LC cells and the external voltage.

Tailoring resonant magnetic perturbation to optimize fast-ion confinement during ELM control in KSTAR

3D resonant magnetic perturbation (RMP) is one promising way to control edge localized modes that can cause excessive material erosion of tokamak first walls. However, RMP can lead to undesired degradation of plasma confinement, including fast-particle losses, which can impact the performance and safety of the reactor. This work investigates the optimization of the poloidal spectrum of the 3D field to optimize fast ion confinement during edge localized mode (ELM) suppression. In the initial step, the validity of the modeling framework is tested against experimental data. Simulations successfully replicate an increase in poloidal limiter temperature with different poloidal spectra. Then, the simulation shows improvement of fast ion confinement with a reduction of core resonant response, while edge resonant magnetic fields are maintained above the threshold to sustain the ELM suppression. Reduction of the core resonant fields keeps the Kolmogorov–Arnold–Moser surface and reduces the fast particle losses due to the stochastic magnetic field lines. The results highlight the potential of edge localization of the resonant fields to enhance the performance of fusion reactors, but further investigation is needed to improve the validation of this approach.

Improving the CFD predictive accuracy of particle fate in turbulent duct flows: Four key issues and applications in in-duct UVGI system

In order to refine the CFD methodology for predicting particle fate in duct flows, this study investigated four key issues largely influencing the predictive accuracy. The flow field modeling method applicable to in-duct UVGI systems (ID-UVs) was also investigated. The particle number independence test indicated that at least 30000 particles should be released. The use of 2D duct cannot consider particle trapping at the sidewalls, thus underestimating the particle deposition behavior. The two-equation k‐ε models significantly overestimated the fluctuation velocity perpendicular to the duct walls in the near-wall region, resulting in them providing Vd+ and η were on average 1767.9 % and 451.3 % higher than those predicted by the RSM model, respectively. Corrections for fluctuation velocity had very limited benefit in improving the predictive accuracy of the two-equation k‐ε model, which unexpectedly resulted in a decrease in the predictive accuracy of the RSM model. Therefore, it was recommended to use the uncorrected RSM model to predict the 3D duct flow field to obtain a high prediction accuracy of particle fate. Similarly, the RSM model was also recommended for modeling the flow field in ID-UVs instead of the two-equation k‐ε model that was often used. This study also revealed that missing the “Wall Reflection Effects” would lead to confusing results and analyzed the reasons for the quantitative differences in the results obtained in previous studies. The findings presented may form a CFD methodological framework for further investigation of the in-duct air purification technologies.

Research and design of Chinese cookie model with chun character based on OpenSCAD

In todays society, the 3D modelling field has become an indispensable part of peoples lives. Whether it is the metaverse, games, industry, or medicine, 3D modelling cannot be separated from them. The models used in various fields are also diverse. In the models of various fields, there are not many models for reference in the food fields biscuit modelling and there are no special views. This article creates a Chinese New Year biscuit with the Chinese handwritten character Spring on it based on the different styles and types of biscuit models in the food industry. OpenSCAD software was used to create this model. This article first introduces the methods and usage of OpenSCAD software that are encapsulated for modelling. Then this paper describes in detail how to layer and create biscuits layer by layer. How to use code to create a handwritten Chinese character and a cream biscuit completely in OpenSCAD is detailed by this article. For the result, this article implements a Chinese New Year biscuit with a reference to the Spring character written by the ancient Chinese calligrapher Yan Zhenqing. This work fully implements every stroke of handwritten Chinese characters in the cookie model, and these details have been recognized by the majority of people in the final evaluation.

Reconstruction and fast prediction of 3D heat and mass transfer based on a variational autoencoder

Reconstruction and fast prediction of heat and mass transfer are important for the improvement of data center operations and energy savings. In this study, an artificial neural network (ANN) and variational autoencoder (VAE) composite model is proposed for the reconstruction and prediction of 3D flowfields with high accuracy and efficiency. The VAE model is trained to extract features of the problem and to realize 3D physical field reconstruction. The ANN is employed to achieve the constructability of the extracted features. A dataset of steady temperature/velocity fields is acquired by computational fluid dynamics and heat transfer (CFD/HT) and fed to train the deep learning model. The proposed ANN-VAE model is experimentally proven to achieve promising field prediction accuracy with a significantly reduced computational cost. CFD/HT method is often incapable of occasions that require instant knowledge of heat and mass transfer. The proposed ANN-VAE method convert the time-consuming simulations into the training stage. Driven by large amount of simulation data, the trained ANN-VAE model is capable of field prediction. Compared to the CFD/HT method, the ANN-VAE method speeds up the physical field prediction by approximately 380,000 times, with mean accuracies of 97.3% for temperature field prediction and 97.9% for velocity field prediction, making it feasible for real-time prediction of heat and mass transfer.

Cross-scale coupling analysis method for OTSG in SFR and its application

The once-through steam generators (OTSG) in sodium-cooled fast reactor (SFR) are characterized by full-regime flow boiling heat transfer, coupled heat transfer on both sides and complex structure with a large dimensional span. This makes the computational fluid dynamics (CFD) and 3D refinement modeling methods to analyze the 3D temperature field of OTSG containing large-scale heat transfer tube bundles generally face the problems of high computational difficulty, low solution efficiency and poor solution stability. In this paper, based on the full-regime flow boiling heat transfer model and the flow pattern discrimination model, a one-dimensional full-regime flow boiling heat transfer model from subcooled water to superheated steam on the water side (tube side) and its cross-scale coupling solution model with the three-dimensional refined CFD model on the sodium side (shell side) are established. The effectiveness of the coupling analysis model was verified by comparative analysis with experimental data of the Indian 19-tube Steam Generator Test Facility (SGTF). The independent coupling of the water side of each heat transfer tube in the OTSG with the sodium side 3D fluid domain is realized. The results show that the cross-scale coupling method using the one-dimensional full-regime flow boiling heat transfer model for the water side and the three-dimensional CFD model for the sodium side can accurately predict the temperature distribution, the main heat transfer characteristic points on the full height of each heat transfer tube, and can obtain the temperature difference between different heat transfer tubes and tube walls. The maximum error between the calculated results and experimental data for the distribution of sodium temperature along the axial direction of the tubes is less than 3%. The cross-scale coupled solution method avoids the direct adoption of the three-dimensional multiphase flow model, which significantly improves the efficiency and stability of the solution of the three-dimensional temperature field of OTSG. This study provides an efficient 3D refinement numerical analysis method for the design and safety verification of OTSG.

MRI-detected intraosseous bone marrow edema recedes after effective therapy of periodontitis

ObjectivesT2 STIR MRI sequences can detect preclinical changes associated with periodontal inflammation, i.e. intraosseous edema in the tooth-supporting bone. In this study, we assessed whether MRI can be used for monitoring periodontal disease.Material and methodsIn a prospective cohort study, we examined 35 patients with periodontitis between 10/2018 and 04/2019 by using 3D isotropic T2-weighted short tau inversion recovery (STIR) and Fast Field Echo T1-weighted Black bone sequences. All patients received standardized clinical exams before and three months after non-surgical periodontal therapy. Bone marrow edema extent was quantified in the STIR sequence at 922 sites before and after treatment. Results were compared with standard clinical findings. Non-parametric statistical analysis was performed.ResultsNon-surgical periodontal treatment caused significant improvement in mean probing depth (p < 0.001) and frequency of bleeding on probing (p < 0.001). The mean depth of osseous edema per site was reduced from a median [IQR] of 2 [1, 3] mm at baseline to 1 [0, 3] mm, (p < 0.001). Periodontal treatment reduced the frequency of sites with edema from 35 to 24% (p < 0.01).ConclusionThe decrease of periodontal bone marrow edema, as observed with T2 STIR MR imaging, is indicative of successful periodontal healing.Clinical relevance statementT2 STIR hyperintense bone marrow edema in the periodontal bone decreases after treatment and can therefore be used to evaluate treatment success. Furthermore, MRI reveals new options to depict hidden aspects of periodontitis.Key Points• T2 STIR hyperintense periodontal intraosseous edema was prospectively investigated in 35 patients with periodontitis before and after treatment and compared to clinical outcomes.• The frequency of affected sites was reduced from 35 to 24% (p < 0.001), and mean edema depth was reduced from a median [IQR] of 2 [1, 3] mm at baseline to 1 [0, 3] mm 3 months after treatment. (p < 0.001).• T2 STIR sequences can be used to monitor the posttreatment course of periodontitis.

Variational quantum metrology for multiparameter estimation under dephasing noise

We present a hybrid quantum-classical variational scheme to enhance precision in quantum metrology. In the scheme, both the initial state and the measurement basis in the quantum part are parameterized and optimized via the classical part. It enables the maximization of information gained about the measured quantity. We discuss specific applications to 3D magnetic field sensing under several dephasing noise models. Indeed, we demonstrate its ability to simultaneously estimate all parameters and surpass the standard quantum limit, making it a powerful tool for metrological applications.

A rapid 3D magnetotelluric forward approach for arbitrary anisotropic conductivities in the Fourier domain

Previous studies have shown that anisotropy generally exists in geological bodies such as sedimentary rocks and fault zones, and more and more attention has been paid to the arbitrary conductivity media in surveys with the magnetotelluric sounding method. With the breakthrough development of computer hardware technology, large-scale 3D magnetotelluric modeling in anisotropic media has gradually become possible. At present, there are 3D magnetotelluric field simulation algorithms based on finite differences or finite elements for arbitrary anisotropic conductivity. In order to solve the common computational efficiency problems of the existing algorithms, we proposed a rapid 3D magnetotelluric forward approach for arbitrary anisotropic conductivity in the Fourier domain. Through the 2D Fourier transform, the governing equation can be converted from the space domain to the Fourier domain, thereby greatly reducing the calculation amount of the numerical simulation and improving the calculation efficiency. Then, the classical 1D anisotropy model is used to verify the correctness and the computational efficiency. Finally, the 3D anisotropic models of land and ocean are calculated, and the influence characteristics of the anisotropic medium on the magnetotelluric response are analyzed. The proposed algorithm will be used in the inverse imaging technique for large-scale 3D anisotropic data in future studies.

In-situ micro-CT damage analysis of carbon and carbon/glass hybrid laminates under tensile loading by image reconstruction and DVC technology

Carbon/ glass fibre-reinforced plastic (C/GFRP) offers advantages in engineering applications due to its low cost. However, the mechanism underlying the evolution of damage in C/GFRP under tension remains unclear. Advanced in-situ micro X-ray computed tomography (µCT) is a useful method for investigating the damage evolution of fibre composites. In this study, two tensile specimens, carbon fibre reinforced plastic (CFRP) and C/GFRP, were prepared, and in-situ µCT scanning was performed at different points during the tensile process. To quantify the voids, examine the fracture morphology, and investigate the void volume fractions, three-dimensional (3D) models were constructed while employing an image reconstruction algorithm. Finally, digital volume correlation (DVC) was employed to study the 3D strain fields. The results suggest that C/GFRP exhibits a higher initial void volume ratio and a faster rate of damage evolution. The 3D strain field and stress concentration area can be used to predict the precise location where the specimen would fracture.

Study on the geothermal environment of urban building in permafrost regions of Northeast China

The geothermal environment of urban buildings is critical to the structural stability of urban buildings in permafrost regions, which is directly related to the safety of human settlements. However, existing studies have not been able to obtain the 3D geothermal field of building subgrades and its evolution during operation, due to a lack of continuous geothermal data. In this study, thermistor sensors were installed inside pile foundations during the construction, providing valuable data for geothermal field analysis. Based on long-term in-situ monitoring data, this paper studied the evolution of the geothermal field of urban building in permafrost regions of Northeast China. The results indicate that excavation and cast-in-place foundation construction cause irreversible degradation of the warm permafrost. When the building is vacant, the subgrade is frozen in the winter, with a maximum frost penetration depth of 6 m. When heated in winter, the frost penetration depth decreases year by year and the ground becomes warmer. The ground temperature within the monitoring depth (i.e., the bearing layer of the pile foundation) increases to above 0 °C. The mean annual ground temperature within the 0–4 m depth of the subgrade increases significantly, while that at depths below 7 m exhibits slight fluctuations. The 3D geothermal field and its evolution of the building subgrade are obtained for the first time. It was also found that the sunny-shady effect induces asymmetry of the geothermal field of the building, which increases the risk of uneven settlement of the building. In addition, the seasonal frost penetration on the northern side reaches 3 m, which can cause frost-heaving damage to buildings. These findings provide scientific evidence for risk assessment, disaster prevention, and control of building in permafrost regions under the scenario of climate warming.

Recent Advances in Photodetectors Based on Two-Dimensional Material/Si Heterojunctions

Two-dimensional (2D) materials have gained significant attention owing to their exceptional electronic and optoelectronic properties, including high carrier mobility, strong light–matter interaction, layer-dependent band structure and band gap. The passivated surface of 2D materials enables the fabrication of van der Waals (vdW) heterojunctions by integrating them with various other materials, such as nanowires, nanosheets and bulk materials. Heterojunction photodetectors, specifically those composed of 2D materials and silicon (Si), have attracted considerable interest due to the well-established processing techniques associated with Si and the excellent performance of the related devices. The hybrid dimension vdW heterojunction composed of 2D materials and Si has the advantages of excellent performance, low fabrication cost, and easy integration with silicon-based devices. It has unique advantages in the field of heterojunction photodetectors. This review provides an overview of the recent advancements in photodetectors based on 2D material/Si heterojunctions. First, we present the background and motivation of the review. Next, we discuss the key performance metrics for evaluating photodetector performance. Then, we review the recent progress made in the field of 2D material/Si heterojunction photodetectors. Finally, we summarize the findings and offer future prospects.

Analyzing 3D fields of refractive index, emission and temperature in spray-flame nanoparticle synthesis via tomographic imaging using multi-simultaneous measurements (TIMes)

Turbulent spray flames issuing from the SpraySyn 2.0 burner were investigated in the present work by means of tomographic imaging using multi-simultaneous measurements (TIMes). This small-scale standard burner is designed to study flame-spray synthesis of tailored nanoparticle systems. The applied measurement technique combined background-oriented schlieren tomography, computed tomography of light emission and two-color tomographic salt emission thermometry to reveal single-shot and time-averaged 3D fields of refractive index (linked to temperature, pressure and species concentration), CH* chemiluminescence (flame front indicator), SrOH emission from synthesized trace amount of Sr(NO3)2 dissolved in the liquid precursor solution (marker of the hot spray-flame products), and temperature of the hot spray-flame products, respectively. The salt emission thermometry technique was extended to tomographic measurements for the first time. Its implementation was based on the voxel-wise intensity ratio of reconstructed thermally-excited SrOH signals at two different regions in the visible spectrum that was converted to temperature using a single calibration curve based on multi-line OH planar laser-induced fluorescence measurements. The cost-effective setup for TIMes was composed of 43 machine-vision cameras equipped with bandpass spectrum filters, an innovative continuous semi-circular background pattern and LED light sources. The combined reconstructed fields provided information on the 3D flame structure in detail, showing the interaction between the flame front and spray stream, and the hot temperature regions. Spray-flame features were quantitatively analyzed, namely temperature, height, width, 3D tilt angle, and spreading half-angle of the flame plume. The temperature measurements proved to be robust against variations in the fuel/oxidizer equivalence ratio for the four investigated operating conditions. The spray flame issuing from the SpraySyn 2.0 burner at the standard operating condition was more stable than that in the previous burner generation, although more asymmetric near the nozzle exit and tilted, which might still cause heterogeneity of the synthesized nanoparticle system.

Simulation of three-dimensional temperature field in high-frequency welding based on nonlinear finite element method

Abstract In modern industrial production, many advanced manufacturing technologies are constantly developing with the progress of social sciences. Welding, as an indispensable manufacturing technology in industrial production, has received close attention from various industries. High frequency welding technology is needed in fields such as mechanical manufacturing, machine making in the food industry, and intelligent robot model making. High frequency welding is an important technical means in the production process of welded pipes, and the level of welding temperature has a significant impact on the quality of welded pipe welds. This article studied the shortcomings of traditional high-frequency welding, analyzed the application method of nonlinear finite element method in high-frequency welding, and analyzed the dynamic process of welding and its influencing factors. The finite element method formula is used to stabilize the value of three-dimensional (3D) temperature field. This work studied the temperature distribution of welded pipe welding, welded pipe materials, inside and outside of welded pipe, and temperature changes under different voltages. The experimental results showed that the error value between the simulation results of the 3D temperature field of high-frequency welding and the measured experimental results was about 4.3542°C, which was basically similar, indicating the effectiveness of the 3D temperature field simulation experiment. With the development of science and technology, high-frequency welding technology would continue to improve, and the quality of welded pipe welds would become better and better with the progress of technology. The improvement in quality promotes the development and progress of industry, and maintains the quality of machine manufacturing. The simulation experiment method of 3D temperature field has shortened the experimental time and reduced the experimental cost, providing a new reference for other temperature related experiments.

3D wind field profiles from hyperspectral sounders: revisiting optic-flow from a meteorological perspective

In this work, we present an efficient optic flow algorithm for the extraction of vertically resolved 3D atmospheric motion vector (AMV) fields from incomplete hyperspectral image data measures by infrared sounders. The model at the heart of the energy to be minimized is consistent with atmospheric dynamics, incorporating ingredients of thermodynamics, hydrostatic equilibrium and statistical turbulence. Modern optimization techniques are deployed to design a low-complexity solver for the energy minimization problem, which is non-convex, non-differentiable, high-dimensional and subject to physical constraints. In particular, taking advantage of the alternate direction of multipliers methods (ADMM), we show how to split the original high-dimensional problem into a recursion involving a set of standard and tractable optic-flow sub-problems. By comparing with the ground truth provided by the operational numerical simulation of the European Centre for Medium-Range Weather Forecasts (ECMWF), we show that the performance of the proposed method is superior to state-of-the-art optical flow algorithms in the context of real infrared atmospheric sounding interferometer (IASI) observations.

4D Flow Patterns and Relative Pressure Distribution in a Left Ventricle Model by Shake-the-Box and Proper Orthogonal Decomposition Analysis

PurposeIntraventricular blood flow dynamics are associated with cardiac function. Accurate, noninvasive, and easy assessments of hemodynamic quantities (such as velocity, vortex, and pressure) could be an important addition to the clinical diagnosis and treatment of heart diseases. However, the complex time-varying flow brings many challenges to the existing noninvasive image-based hemodynamic assessments. The development of reliable techniques and analysis tools is essential for the application of hemodynamic biomarkers in clinical practice.MethodsIn this study, a time-resolved particle tracking method, Shake-the-Box, was applied to reconstruct the flow in a realistic left ventricle (LV) silicone model with biological valves. Based on the obtained velocity, 4D pressure field was calculated using a Poisson equation-based pressure solver. Furthermore, flow analysis by proper orthogonal decomposition (POD) of the 4D velocity field has been performed.ResultsAs a result of the Shake-the-Box algorithm, we have extracted: (i) particle positions, (ii) particle tracks, and finally, (iii) 4D velocity fields. From the latter, the temporal evolution of the 3D pressure field during the full cardiac cycle was obtained. The obtained maximal pressure difference extracted along the base-to-apex was about 2.7 mmHg, which is in good agreement with those reported in vivo. The POD analysis results showed a clear picture of different scale of vortices in the pulsatile LV flow, together with their time-varying information and corresponding kinetic energy content. To reconstruct 95% of the kinetic energy of the LV flow, only the first six POD modes would be required, leading to significant data reduction.ConclusionsThis work demonstrated Shake-the-Box is a promising technique to accurately reconstruct the left ventricle flow field in vitro. The good spatial and temporal resolutions of the velocity measurements enabled a 4D reconstruction of the pressure field in the left ventricle. The application of POD analysis showed its potential in reducing the complexity of the high-resolution left ventricle flow measurements. For future work, image analysis, multi-modality flow assessments, and the development of new flow-derived biomarkers can benefit from fast and data-reducing POD analysis.

Predictive Deep Learning for High‐Dimensional Inverse Modeling of Hydraulic Tomography in Gaussian and Non‐Gaussian Fields

AbstractInverse modeling of hydraulic tomography (HT) is computationally expensive for estimating high‐dimensional hydrogeologic parameter fields. In this work, we develop a novel method called HT‐INV‐NN, which combines dimensionality reduction techniques with a predictive deep learning (DL) model to estimate high‐dimensional Gaussian and non‐Gaussian channel fields. The HT‐INV‐NN model consists of a predictor that directly learns the inverse process from hydraulic head measurements to latent variables of random fields, and a decoder that generates high‐dimensional parameter fields from predicted latent variables. For Gaussian spatially correlated fields, the decoder utilizes principal components derived from spatial covariance, and for non‐Gaussian channel fields, a generative adversarial network (GAN) is trained using generated realizations based on a training image (TI). The predictor is a deep neural network calibrated using the reference data obtained from HT forward simulations, which can be implemented in parallel. HT‐INV‐NN is successfully tested in multiple numerical experiments including steady‐state and transient HT for estimating Gaussian fields in 2D and 3D, as well as binary discontinuous or continuous non‐Gaussian channel fields. The training process is efficient, and the model structure demonstrates robustness for input data with perturbations. The model performance on multiple validation data sets are satisfactory when compared with other numerical and deep learning methods.

Numerical study on the performance of a novel three-dimensional stepped wavy flow field in PEMFC

The design of the flow field plate plays a crucial role in the performance of proton exchange membrane fuel cell (PEMFC). In this study, a novel three-dimensional stepped wavy flow field (SWFF) is developed to enhance the output performance of PEMFC. The effects of geometric parameters of this novel flow field, included the total stepped length (S) and depth (H), the ratio of the sub-stepped length (s1:s2 and s1:s3), and the ratio of the sub-stepped depth (h1:h2), on the transport and distribution characteristics of oxygen and water in PEMFC are numerically investigated. The results demonstrated that the smaller stepped total length and larger stepped total depth result in better cell performance. In addition, the output performance of PEMFC increases with the rise of the ratio of s1:s2, and the decrease of the ratio of s1:s3 under the optimal total stepped length and depth. It is also worth noting that the output performance of PEMFC first increases and then decreases with the rise of the ratio of h1:h2. The optimal values for S, H, s1, s2, s3, h1 and h2 are 8 mm, 0.8 mm, 1.6 mm, 1.6 mm, 1.6 mm, 0.4 mm and 0.4 mm, respectively. Furthermore, when compared to the rectangular baffle flow field, wavy flow field, and parallel flow field, the 3D stepped wavy flow field demonstrated a significant increase in the maximum net output power with improvements of 4.8 %, 11.3 %, and 27.9 %, respectively.

A physics-based machine learning technique rapidly reconstructs the wall-shear stress and pressure fields in coronary arteries.

With the global rise of cardiovascular disease including atherosclerosis, there is a high demand for accurate diagnostic tools that can be used during a short consultation. In view of pathology, abnormal blood flow patterns have been demonstrated to be strong predictors of atherosclerotic lesion incidence, location, progression, and rupture. Prediction of patient-specific blood flow patterns can hence enable fast clinical diagnosis. However, the current state of art for the technique is by employing 3D-imaging-based Computational Fluid Dynamics (CFD). The high computational cost renders these methods impractical. In this work, we present a novel method to expedite the reconstruction of 3D pressure and shear stress fields using a combination of a reduced-order CFD modelling technique together with non-linear regression tools from the Machine Learning (ML) paradigm. Specifically, we develop a proof-of-concept automated pipeline that uses randomised perturbations of an atherosclerotic pig coronary artery to produce a large dataset of unique mesh geometries with variable blood flow. A total of 1,407 geometries were generated from seven reference arteries and were used to simulate blood flow using the CFD solver Abaqus. This CFD dataset was then post-processed using the mesh-domain common-base Proper Orthogonal Decomposition (cPOD) method to obtain Eigen functions and principal coefficients, the latter of which is a product of the individual mesh flow solutions with the POD Eigenvectors. Being a data-reduction method, the POD enables the data to be represented using only the ten most significant modes, which captures cumulatively greater than 95% of variance of flow features due to mesh variations. Next, the node coordinate data of the meshes were embedded in a two-dimensional coordinate system using the t-distributed Stochastic Neighbor Embedding (-SNE) algorithm. The reduced dataset for -SNE coordinates and corresponding vector of POD coefficients were then used to train a Random Forest Regressor (RFR) model. The same methodology was applied to both the volumetric pressure solution and the wall shear stress. The predicted pattern of blood pressure, and shear stress in unseen arterial geometries were compared with the ground truth CFD solutions on "unseen" meshes. The new method was able to reliably reproduce the 3D coronary artery haemodynamics in less than 10 s.

Deep Learning-Based Single-View Fluorescence Dose Reconstruction for 3D Dosimetry

3D dose distribution measurement is crucial for precise radiotherapy. Radiation-excited fluorescence imaging has potential for the 3D dosimetry with high spatial resolution, but multiple fluorescence images from different view-angles are required for analytical reconstruction techniques. Furthermore, the imaging data are contaminated by anisotropic Cherenkov light emission and statistical noise. This project aims to establish a novel deep learning-based model to predict 3D dose distributions from a single-view 2D fluorescence image while simultaneously removing the adverse effects of Cherenkov signals and other noises. A total of 124 single-aperture static photon beams were delivered to an acrylic tank containing 1 g/L quinine hemisulfate water solution with varying aperture shapes and collimator angle. The emitted optical signals were detected by a low-cost CMOS camera for 20 seconds, and image pre-processing was performed to obtain input 2D fluorescence images with 0.3 × 0.3 mm spatial resolution. 3D back-projected dose distribution images were also calculated from the input fluorescence images. Ground-truth of 3D dose distributions and 2D field map images were obtained from a clinical treatment planning system with 1.4 × 1.4 × 1.4 mm spatial resolution. The proposed deep learning-based dose reconstruction method involved 3 steps. First, 2D fluence map images at the bottom plane of the tank were predicted from the fluorescence images by using a customized convolutional neural network (CNN). Second, the predicted fluence map images were transformed into the 2D field map images on the isocenter plane by applying perspective transformation. Finally, 2D dose distributions at a given radiological depth were calculated by using the predicted field map images, the back-projected dose distribution images, and the radiological depth value as inputs of a shallow CNN. Both CNN models were trained separately, and the 3D dose distributions were predicted by concatenating the output 2D dose distributions at various radiological depths. The proposed CNN model yielded accurate 2D field map images. Averaged Dice similarity coefficient and mean absolute error of the field maps in the test data was 92.0% ± 4.6% and 0.0132 ± 0.0113, respectively. Moreover, our deep learning-based approach was able to predict accurate 3D dose distributions from the 2D fluorescence images. Mean squared error and averaged 3D gamma passing ratio (3%/3mm) were 9.55 mGy ± 6.8 mGy and 86.3% ± 9.86%, respectively. Theproposed deep learning-based method calculated accurate 3D dose distributions from a single-view 2D fluorescence image. Since this technique require only a single CMOS camera image and fluorescent material, it can be readily used for any external radiotherapy modalities, including SRS/SBRT with small fields. This method is useful for acquiring 3D dose distribution data for precise dose verification within a few seconds.