Data-driven continuum damage mechanics with built-in physics.

Soft materials such as rubbers and soft tissues often undergo large deformations and experience damage degradation that impairs their function. This energy dissipation mechanism can be described in a thermodynamically consistent framework known as continuum damage mechanics. Recently, data-driven methods have been developed to capture complex material behaviors with unmatched accuracy due to the high flexibility of deep learning architectures. Initial efforts focused on hyperelastic materials, and recent advances now offer the ability to satisfy physics constraints such as polyconvexity of the strain energy density function by default. However, modeling inelastic behavior with deep learning architectures and built-in physics has remained challenging. Here we show that neural ordinary differential equations (NODEs), which we used previously to model arbitrary hyperelastic materials with automatic polyconvexity, can be extended to model energy dissipation in a thermodynamically consistent way by introducing an inelastic potential: a monotonic yield function. We demonstrate the inherent flexibility of our network architecture in terms of different damage models proposed in the literature. Our results suggest that our NODEs re-discover the true damage function from synthetic stress-deformation history data. In addition, they can accurately characterize experimental skin and subcutaneous tissue data.

Physics-Informed Machine Learning for Accurate Prediction of Temperature and Melt Pool Dimension in Metal Additive Manufacturing.

The temperature distribution and melt pool size have a great influence on the microstructure and mechanical behavior of metal additive manufacturing process. The numerical method can give relatively accurate results but is time-consuming and, therefore, unsuitable for in-process prediction. Owing to its remarkable capabilities, machine learning methods have been applied to predict melt pool size and temperature distribution. However, the success of traditional data-driven machine learning methods is highly dependent on the amount and quality of the training data, which is not always convenient to access. This article proposes a physics-informed machine learning (PIML) method, which integrates data and physics laws in the training parts, overcoming the problems of low speed and data availability. An artificial neural network constrained by the heat transfer equation and a small amount of labeled data is developed to predict the melt pool size and temperature distribution. Besides, the locally adaptive activation function is utilized to improve the prediction performance. The result shows that the developed PIML model can accurately predict the temperature and melt pool dimension under different scanning speeds with a small amount of labeled data, which shows significant potential in practical application.

CT perfusion parameter estimation in stroke using neural network with transformer and physical model priors.

OBJECTIVES: CT perfusion (CTP) imaging is vital in treating acute ischemic stroke by identifying salvageable tissue and the infarcted core. CTP images allow quantitative estimation of CT perfusion parameters, which can provide information on the degree of tissue hypoperfusion and its salvage potential. Traditional methods for estimating perfusion parameters, such as singular value decomposition (SVD) and its variations, are known to be sensitive to noise and inaccuracies in the arterial input function. To our knowledge, there has been no implementation of deep learning methods for CT perfusion parameter estimation. MATERIALS & METHODS: In this work, we propose a deep learning method based on the Transformer model, named CTPerformer-Net, for CT perfusion parameter estimation. In addition, our method incorporates some physical priors. We integrate physical consistency prior, smoothness prior and the physical model prior through the design of the loss function. We also generate a simulation dataset based on physical model prior for training the network model. RESULTS: In the simulation dataset, CTPerformer-Net exhibits a 23.4 % increase in correlation coefficients, a 95.2 % decrease in system error, and a 90.7 % reduction in random error when contrasted with block-circulant SVD. CTPerformer-Net successfully identifies hypoperfused and infarcted lesions in 103 real CTP images from the ISLES 2018 challenge dataset. It achieves a mean dice score of 0.36 for the infarct core segmentation, which is slightly higher than the commercially available software (dice coefficient: 0.34) used as a reference level by the challenge. CONCLUSION: Experimental results on the simulation dataset demonstrate that CTPerformer-Net achieves better performance compared to block-circulant SVD. The real-world patient dataset confirms the validity of CTPerformer-Net.

Local volume concentration, packing domains, and scaling properties of chromatin.

We propose the Self Returning Excluded Volume (SR-EV) model for the structure of chromatin based on stochastic rules and physical interactions. The SR-EV rules of return generate conformationally defined domains observed by single-cell imaging techniques. From nucleosome to chromosome scales, the model captures the overall chromatin organization as a corrugated system, with dense and dilute regions alternating in a manner that resembles the mixing of two disordered bi-continuous phases. This particular organizational topology is a consequence of the multiplicity of interactions and processes occurring in the nuclei, and mimicked by the proposed return rules. Single configuration properties and ensemble averages show a robust agreement between theoretical and experimental results including chromatin volume concentration, contact probability, packing domain identification and size characterization, and packing scaling behavior. Model and experimental results suggest that there is an inherent chromatin organization regardless of the cell character and resistant to an external forcing such as RAD21 degradation.

Polymer Physics Models Reveal Structural Folding Features of Single-Molecule Gene Chromatin Conformations.

Here, we employ polymer physics models of chromatin to investigate the 3D folding of a 2 Mb wide genomic region encompassing the human LTN1 gene, a crucial DNA locus involved in key cellular functions. Through extensive Molecular Dynamics simulations, we reconstruct in silico the ensemble of single-molecule LTN1 3D structures, which we benchmark against recent in situ Hi-C 2.0 data. The model-derived single molecules are then used to predict structural folding features at the single-cell level, providing testable predictions for super-resolution microscopy experiments.

Non-equilibrium formation and relaxation of magnetic flux ropes at kinetic scales.

Magnetic flux ropes are pivotal structures and building blocks in astrophysical and laboratory plasmas, and various equilibrium models have thus been studied in the past. However, flux ropes in general form at non-equilibrium, and their pathway from formation to relaxation is a crucial process that determines their eventual properties. Here we show that any localized current parallel to a background magnetic field will evolve into a flux rope via non-equilibrium processes. The detailed kinetic dynamics are exhaustively explained through single-particle and Vlasov analyses and verified through particle-in-cell simulations. This process is consistent with many proposed mechanisms of flux rope generation such as magnetic reconnection. A spacecraft observation of an example flux rope is also presented; by invoking the non-equilibrium process, its structure and properties can be explicated down to all six components of the temperature tensor.

Reforming Medical Physics and Radiopharmaceutical Science Training Through a Programmatic Approach to Assessment.

OBJECTIVES: Programmatic assessment approaches can be extended to the design of allied health professions training, to enhance the learning of trainees. The Australasian College of Physical Scientists and Engineers in Medicine worked with assessment specialists at the Australian Council for Educational Research and Amplexa Consulting, to revise their medical physics and radiopharmaceutical science training programs. One of the central aims of the revisions was to produce a training program that provides standardized training support to their registrars throughout the 3 years, better supporting their registrars to successfully complete the program in the time frame through providing timely and constructive feedback on the registrar's progression. METHODS: We used the principles of programmatic assessment to revise the assessment methods and progression decisions in the three training programs. RESULTS: We revised the 3-year training programs for diagnostic imaging medical physics, radiation oncology medical physics and radiopharmaceutical science in Australia and New Zealand, incorporating clear stages of training and associated progression points. CONCLUSIONS: We discuss the advantages and difficulties that have arisen with this implementation. We found 5 key elements necessary for implementing programmatic assessment in these specialized contexts: embracing blurred boundaries between assessment of and for learning, adapting the approach to each specialized context, change management, engaging subject matter experts, and clear communication to registrars/trainees.

Super-Resolving and Denoising 4D flow MRI of Neurofluids Using Physics-Guided Neural Networks.

PURPOSE: To obtain high-resolution velocity fields of cerebrospinal fluid (CSF) and cerebral blood flow by applying a physics-guided neural network (div-mDCSRN-Flow) to 4D flow MRI. METHODS: The div-mDCSRN-Flow network was developed to improve spatial resolution and denoise 4D flow MRI. The network was trained with patches of paired high-resolution and low-resolution synthetic 4D flow MRI data derived from computational fluid dynamic simulations of CSF flow within the cerebral ventricles of five healthy cases and five Alzheimer's disease cases. The loss function combined mean squared error with a binary cross-entropy term for segmentation and a divergence-based regularization term for the conservation of mass. Performance was assessed using synthetic 4D flow MRI in one healthy and one Alzheimer' disease cases, an in vitro study of healthy cerebral ventricles, and in vivo 4D flow imaging of CSF as well as flow in arterial and venous blood vessels. Comparison was performed to trilinear interpolation, divergence-free radial basis functions, divergence-free wavelets, 4DFlowNet, and our network without divergence constraints. RESULTS: The proposed network div-mDCSRN-Flow outperformed other methods in reconstructing high-resolution velocity fields from synthetic 4D flow MRI in healthy and AD cases. The div-mDCSRN-Flow network reduced error by 22.5% relative to linear interpolation for in vitro core voxels and by 49.5% in edge voxels. CONCLUSION: The results demonstrate generalizability of our 4D flow MRI super-resolution and denoising approach due to network training using flow patches and physics-based constraints. The mDCSRN-Flow network can facilitate MRI studies involving CSF flow measurements in cerebral ventricles and association of MRI-based flow metrics with cerebrovascular health.

Comparison of compartmental analytical Blood-Oxygen-Level-Dependent functional Magnetic Resonance Imaging models against Monte Carlo simulations performed over cortical micro-angiograms.

Blood oxygen level-dependent functional magnetic resonance imaging (BOLD fMRI) arises from a physiological and physical cascade of events taking place at the level of the cortical microvasculature which constitutes a medium with complex geometry. Several analytical models of the BOLD contrast have been developed, but these have not been compared directly against detailed bottom-up modeling methods. Using a 3D modeling method based on experimentally measured images of mice microvasculature and Monte Carlo simulations, we quantified the accuracy of two analytical models to predict the amplitude of the BOLD response from 1.5 to 7 T, for different echo time (TE) and for both gradient echo and spin echo acquisition protocols. We also showed that accounting for the tridimensional structure of the microvasculature results in more accurate prediction of the BOLD amplitude, even if the values for SO2 were averaged across individual vascular compartments. A secondary finding is that modeling the venous compartment as two individual compartments results in more accurate prediction of the BOLD amplitude compared with standard homogenous venous modeling, arising from the bimodal distribution of venous SO2 across the microvasculature in our data.

Visual detection and enhancement of alterations in questioned documents by Mueller polarimetry.

In this article, we have explored the use of Mueller polarimetry for the detection and enhancement of alterations in questioned documents. Erasures, obliterations (with liquid paper and by pasting an additional layer of paper), and insertions (made with several inks) were studied in both regular and glossy paper. Promising results were obtained, which depend on the type of paper and the relation between the color of the ink and that of the illuminating light source. Erasures are easier to detect in glossy paper than in regular paper. Obliterations with liquid paper produced successful results in both types of paper, while detection of obliterations made with an additional layer of paper led to higher contrast for regular paper. Regarding the insertions, the black ball-point ink could be differentiated from roller-ball and gel-pen ink, which is often difficult to achieve visually. The contrast observed between the two inks was higher for regular paper than for glossy paper. Although the results shown in this article are promising, a wider variety of papers and pen types must be tested to further develop the procedure. It has the advantage of being non-destructive and far more economic than other methods. In some cases, the results can be complementary to those obtained by other methods (e.g., fluorescence with UV excitation and illumination with transmitted and oblique light), while in other cases the method offers unique advantages.

Modularized neural network incorporating physical priors for future building energy modeling.

Building energy modeling (BEM) is fundamental for achieving optimized energy control, resilient retrofit designs, and sustainable urbanization to mitigate climate change. However, traditional BEM requires detailed building information, expert knowledge, substantial modeling efforts, and customized case-by-case calibrations. This process must be repeated for every building, thereby limiting its scalability. To address these limitations, we developed a modularized neural network incorporating physical priors (ModNN), which is improved by its model structure incorporating heat balance equations, physically consistent model constraints, and data-driven modular design that can allow for multiple-building applications through model sharing and inheritance. We demonstrated its scalability in four cases: load prediction, indoor environment modeling, building retrofitting, and energy optimization. This approach provides guidance for future BEM by incorporating physical priors into data-driven models without extensive modeling efforts, paving the way for large-scale BEM, energy management, retrofit designs, and buildings-to-grid integration.

Toward the Integration of Machine Learning and Molecular Modeling for Designing Drug Delivery Nanocarriers.

The pioneering work on liposomes in the 1960s and subsequent research in controlled drug release systems significantly advances the development of nanocarriers (NCs) for drug delivery. This field is evolved to include a diverse array of nanocarriers such as liposomes, polymeric nanoparticles, dendrimers, and more, each tailored to specific therapeutic applications. Despite significant achievements, the clinical translation of nanocarriers is limited, primarily due to the low efficiency of drug delivery and an incomplete understanding of nanocarrier interactions with biological systems. Addressing these challenges requires interdisciplinary collaboration and a deep understanding of the nano-bio interface. To enhance nanocarrier design, scientists employ both physics-based and data-driven models. Physics-based models provide detailed insights into chemical reactions and interactions at atomic and molecular scales, while data-driven models leverage machine learning to analyze large datasets and uncover hidden mechanisms. The integration of these models presents challenges such as harmonizing different modeling approaches and ensuring model validation and generalization across biological systems. However, this integration is crucial for developing effective and targeted nanocarrier systems. By integrating these approaches with enhanced data infrastructure, explainable AI, computational advances, and machine learning potentials, researchers can develop innovative nanomedicine solutions, ultimately improving therapeutic outcomes.

Female-dominated disciplines have lower evaluated research quality and funding success rates, for men and women.

We use data from 30 countries and find that the more women in a discipline, the lower quality the research in that discipline is evaluated to be and the lower the funding success rate is. This affects men and women, and is robust to age, number of research outputs, and bibliometric measures where such data are available. Our work builds on others' findings that women's work is valued less, regardless of who performs that work.

There have been growing concerns around sexism in science. Studies have found that women in science are often paid less, are less likely to get credit for their work and receive fewer and smaller grants than men at similar stages in their careers. This can make it harder for women to advance in their careers, resulting in less women than men taking up positions of leadership. There are also gender imbalances between scientific disciplines, with a higher proportion of women working in some fields compared to others. Here, James et al. set out to find whether having more women working in a discipline leads to biases in how the research is evaluated. The team examined four datasets which included information on the research evaluations and funding success of thousands of researchers across 30 different countries. The analysis suggested that scientists working in women-dominated disciplines were less likely to succeed in their grant applications. Their research was also often evaluated as being lower quality compared to researchers working in fields dominated by men. These biases applied to both men and women working in these disciplines. There were not sufficient data to analyse patterns faced by non-binary individuals. The study by James et al. cannot pinpoint a specific cause for these outcomes. However, it suggests that funding organisations should analyse the pattern of successful applications across disciplines and consider taking steps to ensure all disciplines have similar success rates. James et al. also propose that when hiring or making promotions, scientific institutions should take care when comparing researchers across disciplines and ensure there is no built-in assumption that fields dominated by men are intrinsically better.

Optical characteristics of laser medical instrument with side-firing fiber under complete bevel angle range.

The side-firing instrument is studied under the complete bevel angle range. The fiber core and cladding are 0.6 mm and 0.66 mm, respectively, the fiber core refractive index is 1.457, and the fiber cladding refractive index is from 1.409 to 1.452 corresponding to the numerical aperture from 0.37 to 0.12. The bevel angle range is subdivided by ten crucial angles, whose relationship changes as the fiber cladding refractive index reaches 1.418. The beam's divergence angle and coverage increase as the bevel angle deviates from being equal and close to π/4 rad, respectively. When all rays achieve total internal reflection, with numerical aperture being 0.37, the divergence angle and coverage would increase by 28.29% and 44.74%, respectively. The required emission opening size has a minimum under the bevel angle being close to π/4 rad, whose expression is obtained. It increases sharply as the bevel angle reaches a certain value.