Deformation Framework Research Articles

A Real‐Time Imaging Sensing System to Visualize Elastomer Surface Profile Evolution for Dynamic Tactile Recognition

AbstractElastomer‐based interfaces provide rich functionalities for tactile sensing, particularly in making tiny differences in contact dynamics potentially detectable. However, the minimal motion‐induced changes in the elastomer's surface and their disappearance during time‐lapse limit the state recognition within the current scheme of motion recognition. In this work, a new scheme of real‐time motion mode recognition for subtle deformations is proposed, which uses an optical tactile sensing system to visualize and distinguish tiny variations in surface profile evolution encoded as images. Illustrating with a sphere, sliding‐induced asymmetric elastomer surface deformation is visualized as a “drag” in optical images. The convolutional neural network (CNN) algorithm is used to analyze the evolution of surface contour features during the interaction between the sphere and the elastic medium. Motion state recognition is achieved with 80% accuracy when a displacement of only 8.3% of the sphere diameter is produced. In addition, the system also offers the potential to analyze dynamic motion information through a single image, with an accuracy of 82.7% for velocity recognition. This dynamic real‐time recognition framework for soft media deformation paves the way for novel motion‐based input commands for tactile sensing and human‐computer interaction applications.

A new adaptive peridynamic framework for modeling large deformation and fracture behavior of hyperelastic materials

Accurate prediction of deformation and fracture behaviors of hyperelastic materials is crucial in engineering applications and industry. The paper presents a new three-dimensional computational model to simulate large deformations and fractures in hyperelastic materials using the peridynamic method. The main innovation of the proposed three-dimensional hyperelastic peridynamic (3D-HPD) model lies in its incorporation of the stress–strain relationship of hyperelastic materials into the peridynamic framework. Therefore, the peridynamic parameters, such as bond stiffness and strength, are dynamically updated based on the instantaneous value of Young's modulus derived from the stress–strain relation of hyperelastic materials. The 3D-HPD model highlights the following novelties: 1) The 3D-HPD model directly employs the stress–strain relation for each peridynamic bond. Thus, the large deformations and fractures in hyperelastic materials are accurately captured in three dimensions by integrating the responses in all peridynamic bonds; 2) Unlike other models that may require specific hyperelastic formulations, the 3D-HPD model offers a more straightforward application, enhancing usability and reducing complexity; 3) The 3D-HPD model is capable of simulating the response of hyperelastic materials under various loading conditions, including tensile, torsional, and lateral loadings. We confirm the validity of the 3D-HPD model by evaluating it against experimental measurements and observations. For the first time, the model examines hyperelastic materials' fracture and deformation behaviors under combined loading conditions. The new findings of the mechanical performance of hyperelastic materials under multi-axial stresses provide new insights to enhance their design and usage.

Local stress/strain field analysis of die-casting Al alloys via 3D model simulation with realistic defect distribution and RVE modelling

The deformation and fracture behavior of die-casting aluminium (Al) alloys is very complex. Due to local variations in properties, the microstructure and mechanical behavior of materials are highly anisotropic. In this paper, an attempt has been made to quantitatively study the defect characteristics of high pressure cast Al alloy parts using experimental and finite element calculation methods, and to analysis the effect of local porosity and pore size on plasticity. Three-dimensional solids with real defect distribution are obtained by using 3D X-ray computed tomography, and used as an input for building a finite element model. The damage initiation of cast Al alloys under complex stress states is analyzed from micro to macro scales. Crack propagation occurs through two modes of microporous agglomeration: aggregated pores produce cracks from internal necking and stress concentration. Then, they expand in the same direction, agglomerate in a specific direction and eventually fracture. Subsequently, the effect of porosity on non-homogeneity was elucidated by obtaining local stress/strain behavior through digital image correlation measurements. Additionally, a theoretical framework of elasto-plastic deformation of microstructures and a 3D representative volume element model are developed to simulate the deformation and damage processes under cyclic boundary conditions of materials. The simulation results show that the local stress/strain around the pores evolves gradually with deformation. During the die casting process, the method demonstrates the ability to predict the mechanical behavior of Al alloys.

Enhancing the Aesthetics of 3D Shapes via Reference-based Editing

While there have been previous works that explored methods to enhance the aesthetics of images, the automated beautification of 3D shapes has been limited to specific shapes such as 3D face models. In this paper, we introduce a framework to automatically enhance the aesthetics of general 3D shapes. Our approach employs a reference-based beautification strategy. We first performed data collection to gather the aesthetics ratings of various 3D shapes to create a 3D shape aesthetics dataset. Then we perform reference-based editing to edit the input shape and beautify it by making it look more like some reference shape that is aesthetic. Specifically, we propose a reference-guided global deformation framework to coherently deform the input shape such that its structural proportions will be closer to those of the reference shape. We then optionally transplant some local aesthetic parts from the reference to the input to obtain the beautified output shapes. Comparisons show that our reference-guided 3D deformation algorithm outperforms existing techniques. Furthermore, quantitative and qualitative evaluations demonstrate that the performance of our aesthetics enhancement framework is consistent with both human perception and existing 3D shape aesthetics assessment.

A physics knowledge-based surrogate model framework for time-dependent slope deformation: Considering water effect and sliding states

The surrogate model serves as an efficient simulation tool during the slope parameter inversion process. However, the creep constitutive model integrated with dynamic damage evolution poses challenges in development of the required surrogate model. In this study, a novel physics knowledge-based surrogate model framework is proposed. In this framework, a Transformer module is employed to capture strain-driven softening-hardening physical mechanisms. Positional encoding and self-attention are utilized to transform the constitutive parameters associated with shear strain, which are not directly time-related, into intermediate latent features for physical loss calculation. Next, a multi-layer stacked GRU (gated recurrent unit) network is built to provide input interfaces for time-dependent intermediate latent features, hydraulic boundary conditions, and water-rock interaction degradation equations, with static parameters introduced via external fully-connected layers. Finally, a combined loss function is constructed to facilitate the collaborative training of physical and data loss, introducing time-dependent weight adjustments to focus the surrogate model on accurate deformation predictions during critical phases. Based on the deformation of a reservoir bank landslide triggered by impoundment and subsequent restabilization, an elasto-viscoplastic constitutive model that considers water effect and sliding state dependencies is developed to validate the proposed surrogate model framework. The results indicate that the framework exhibits good performance in capturing physical mechanisms and predicting creep behavior, reducing errors by about 30 times compared to baseline models such as GRU and LSTM (long short-term memory), meeting the precision requirements for parameter inversion. Ablation experiments also confirmed the effectiveness of the framework. This framework can also serve as a reference for constructing other creep surrogate models that involve non-time-related across dimensions.

Asymmetric Deformable Spatio-temporal Framework for Infrared Object Tracking

The Infrared Object Tracking (IOT) task aims to locate objects in infrared sequences. Since color and texture information is unavailable in infrared modality, most existing infrared trackers merely rely on capturing spatial contexts from the image to enhance feature representation, where other complementary information is rarely deployed. To fill this gap, we in this article propose a novel Asymmetric Deformable Spatio-Temporal Framework (ADSF) to fully exploit collaborative shape and temporal clues in terms of the objects. Firstly, an asymmetric deformable cross-attention module is designed to extract shape information, which attends to the deformable correlations between distinct frames in an asymmetric manner. Secondly, a spatio-temporal tracking framework is coined to learn the temporal variance trend of the object during the training process and store the template information closest to the tracking frame when testing. Comprehensive experiments demonstrate that ADSF outperforms state-of-the-art methods on three public datasets. Extensive ablation experiments further confirm the effectiveness of each component in ADSF. Furthermore, we conduct generalization validation to demonstrate that the proposed method also achieves performance gains in RGB-based tracking scenarios.

Correction method and verification of radial inertia and friction effects under a unified deformation framework in SHPB experiments on soft materials

During the split Hopkinson pressure bar (SHPB) experiments, significant measurement errors can arise due to severe radial inertia and friction effects. Previous studies have developed various correction methods for these two effects. However, these methods have problems such as over-reliance on the volume invariance assumption of the specimen and inconsistent assumptions on the deformation patterns of the two effects, which limit their universality and effectiveness. Therefore, this paper integrates the radial inertia effect and friction effect in a unified deformation framework through reasonable assumptions, and proposes a method to correct the specimen from a complex stress state to a uniaxial stress state. SHPB numerical simulation experiments demonstrate that this method effectively eliminates the combined effects of radial inertia and friction on measurement results for both elastic and viscoelastic materials, including the size effect associated with these two factors. Additionally, the paper presents a scheme to determine the friction coefficient using the size effect of the specimens when the friction coefficient between the specimen and the bar is unknown. Finally, the method was applied to correct the stresses measured in SHPB experiments on silicone rubber of different diameters. It successfully eliminated discrepancies in the stress-strain relationships between specimens of various sizes and determined a friction coefficient that fell within a reasonable range.

Detailed particle breakage of crushable granular materials under cyclic shearing conditions using dyed gypsum technique

Particle breakage of granular materials under cyclic shearing is related to a variety of engineering problems in geotechnical and transportation engineering. However, there is limited understanding regarding the detailed evolution law of breakage under cyclic shearing. To this end, a series of cyclic simple shear tests were conducted on artificially dyed gypsum particles. It was observed that, in contrast to the particle size distribution of the whole sample, the fractional particle size distributions of gap-graded samples followed a unified breakage evolution path as those of the uniformly graded ones and tended towards fractional fractal distributions. These results have inspired the introduction of the fractional breakage index. Moreover, the breakage-plastic work relationship was further extended to describe the breakage of fractional particles, incorporating the effect of the number of cycles on the plastic work distribution. Finally, based on the concept of breakage-packing, a predictive model for plastic work-breakage-deformation of crushable granular materials under cyclic shearing was proposed. These results have the potential in understanding the detailed particle breakage evolution and establishing a predictive framework for the breakage-induced deformation of crushable granular materials.

Semantic-aware hyper-space deformable neural radiance fields for facial avatar reconstruction

High-fidelity facial avatar reconstruction from monocular videos is a prominent research problem in computer graphics and computer vision. Recent advancements in the Neural Radiance Field (NeRF) have demonstrated remarkable proficiency in rendering novel views and garnered attention for its potential in facial avatar reconstruction. However, previous methodologies have overlooked the complex motion dynamics present across the head, torso, and intricate facial features. Additionally, a deficiency exists in a generalized NeRF-based framework for facial avatar reconstruction adaptable to either 3DMM coefficients or audio input. To tackle these challenges, we propose an innovative framework that leverages semantic-aware hyper-space deformable NeRF, facilitating the reconstruction of high-fidelity facial avatars from either 3DMM coefficients or audio features. Our framework effectively addresses both localized facial movements and broader head and torso motions through semantic guidance and a unified hyper-space deformation module. Specifically, we adopt a dynamic weighted ray sampling strategy to allocate varying degrees of attention to distinct semantic regions, enhancing the deformable NeRF framework with semantic guidance to capture fine-grained details across diverse facial regions. Moreover, we introduce a hyper-space deformation module that enables the transformation of observation space coordinates into canonical hyper-space coordinates, allowing for the learning of natural facial deformation and head-torso movements. Extensive experiments validate the superiority of our framework over existing state-of-the-art methods, demonstrating its effectiveness in producing realistic and expressive facial avatars. Our code is available at https://github.com/jematy/SAHS-Deformable-Nerf.

Reducing Motion Artifacts in Craniocervical Background Subtraction Angiography with Deformable Registration and Unsupervised Deep Learning

Abstract Background In clinical practice, digital subtraction angiography (DSA) often suffers from misregistration artifact due to voluntary, respiratory, and cardiac motion during acquisition. Most prior efforts to register the background DSA mask to subsequent post-contrast images rely on key point registration using iterative optimization, which has limited real-time application. Purpose Leveraging state-of-the-art, unsupervised deep learning, we aim to develop a fast, deformable registration model to substantially reduce DSA misregistration in craniocervical angiography without compromising spatial resolution or introducing new artifacts. Methods We extend HyperMorph, an open source deep learning deformable registration framework, to reduce motion artifacts in DSA. Novel image similarity loss functions with vessel layer estimation were introduced to optimize background registration, making it robust to the variable presence of intravascular iodinated contrast. Results A total of 516 studies with 5,240 angiographic series were collected and divided into training (5046 series) and hold-out test (194 series) sets. Blinded algorithm rankings and Likert scores on five-point scales (1 = worst, 5 = best) were generated by three practicing interventional neuroradiologists using 50 series randomly selected from the hold-out test set. Compared to traditional DSA, our learning-based background subtraction angiography (BSA) significant improved vascular fidelity (2.4 ± 0.6 for DSA vs. 3.6 ± 0.5 for BSA), subtraction artifacts (2.0 ± 0.4 for DSA vs. 3.9 ± 0.3 for BSA), and overall quality (2.1 ± 0.5 for DSA vs. 3.9 ± 0.4 for BSA) (p < 0.0001). Learning-based BSA also significantly outperformed affine registration-based BSA (p < 0.0001). The average inference time for learning-based BSA was 30 milliseconds per frame on our hardware. Conclusion The results demonstrate that deep learning deformable registration, combined with an appropriate loss function, can significantly reduce the motion artifacts that degrade DSA.

Predicting buckling regimes during first lithiation of a crystalline silicon cylindrical anode particle: Influence of size and charging rate

A mathematical model for buckling analysis during the first lithiation of cylindrical crystalline silicon (c-Si) anode particles is presented using the finite deformation framework. A reaction–diffusion model that incorporates a reversible alloying–dealloying reaction (ADR) captures the lithiation in the expanding amorphous zone, while an addition reaction models the dynamics of the crystalline-amorphous silicon interface. A “movable” lithium part and an “immovable” lithium part make up the full lithium. In sharp contrast to the case of amorphous silicon (a-Si), the axial force evolution in c-Si is found to show two different peaks during the initial and final stages of lithiation, respectively. These peaks form the basis of two different buckling criteria, which, in turn, are shown to be sensitively influenced by the influx of lithium and a crucial size-dependent parameter. This framework gives an insight towards investigating the buckling phenomena for lithiation with an evolving amorphous silicon zone due to the interface movement, which is still absent in the present literature. Additionally, the present model enables us to obtain a quantitative and mechanistic insight into and capture the previous experimental observation that the mechanical stability of a cylindrical silicon particle may be improved by maintaining a crystalline silicon core inside the amorphous shell through controlled first lithiation. It is shown that such a crystalline-amorphous core–shell structure can be advantageous to more complicated designs involving carbon fiber as the core. It is thus hoped that this model will contribute towards improving the theoretical underpinnings of an improved design of mechanically robust electrodes for next-generation batteries.

Non-Contact Blood Pressure Estimation From Radar Signals by a Stacked Deformable Convolution Network.

This study introduces a contactless blood pressure monitoring approach that combines conventional radar signal processing with novel deep learning architectures. During the preprocessing phase, datasets suitable for synchronization are created by integrating Kalman filtering, multiscale bandpass filters, and a periodic extraction method in the time domain. These data comprise data on chest micro variations, encapsulating a complex array of physiological and biomedical information reflective of cardiac micromotions. The Radar-based Stacked Deformable convolution Network (RSD-Net) integrates channel and spatial self attention mechanisms within a deformable convolutional framework to enhance feature extraction from radar signals. The network architecture systematically employs deformable convolutions for initial deep feature extraction from individual signals. Subsequently, continuous blood pressure estimation is conducted using self attention mechanisms on feature map from single source coupled with multi-feature map channel attention. The performance of model is corroborated via the open-source dataset procured using a non-invasive 24 GHz six-port continuous wave radar system. The dataset, encompassing readings from 30 healthy individuals subjected to diverse conditions including rest, the Valsalva maneuver, apnea, and tilt-table examinations. It serves to substantiate the validity and resilience of the proposed method in the non-contact assessment of continuous blood pressure. Evaluation metrics reveal Pearson correlation coefficients of 0.838 for systolic and 0.797 for diastolic blood pressure predictions. The Mean Error (ME) and Standard Deviation (SD) for systolic and diastolic blood pressure measurements are -0.32 ±6.14 mmHg and -0.20 ±5.50 mmHg, respectively. The ablation study assesses the contribution of different structural components of the RSD-Net, validating their significance in the overall of model performance.

A three-dimensional finite strain volumetric cohesive XFEM-based model for ductile fracture

The verification/prediction by numerical simulation of the resistance of large-scale metal structures subjected to extreme loading (e.g. accidental mechanical overload) that can lead to failure constitutes a major issue in the transport, energy, and defense sectors. This work aims to develop a three-dimensional numerical methodology (i) capable of macroscopically accounting for the various dissipative mechanisms of plasticity and damage until failure, (ii) formulated within the framework of large deformation, (iii) implemented in a commercial computation code, and (iv) mesh-objective. The commercial computational code is ABAQUS-std, and the methodology is implemented as a user finite element (UEL). Geometric non-linearities are treated within the framework of an updated Lagrangian formulation (ULF). The more or less diffuse damage phase is described by the Gurson–Tvergaard–Needleman (GTN) microporous plasticity model with standard finite element (FE). The micro-void coalescence phase-induced dilation/shear band is described using an original volumetric cohesive extended finite element method approach (VCZM-XFEM). The progressive loss of cohesion leading to ultimate cracking is treated with extended finite element method (XFEM). Particular attention is paid to transition criteria (damage to localization, localization to cracking), to the orientation of the localization plane (especially in Mode II), to the cohesive zone model and to techniques for overcoming numerical difficulties (volumetric locking, numerical integration). The so-built 3D-FS-GTN-VCZM-XFEM unified methodology is capable of reproducing the degradation mechanisms, and, the failure surface shape and orientation in large elasto-plastic deformation of structures typical of laboratory tests, even for fairly coarse meshes.

Quantum van Est isomorphism

Motivated by the fact that the Hopf-cyclic (co)homologies of function algebras over Lie groups and universal enveloping algebras over Lie algebras capture the Lie group and Lie algebra (co)homologies, we hereby upgrade the classical van Est isomorphism to ones between the Hopf-cyclic (co)homologies of quantized algebras of functions and quantized universal enveloping algebras, both in h -adic and q -deformation frameworks.

Geochronology of the Paleoproterozoic Helanshan ductile shear zones: Insights into temporal framework of polyphase deformation in the Khondalite Belt, North China Craton

The Khondalite Belt has been regarded as a Paleoproterozoic collisional orogen in the northwestern North China Craton, but the geochronological framework of polyphase deformation in the Khondalite Belt still remains unclear, particularly the timing of orogen-parallel ductile shearing. In this study, we conducted field-based structural analysis and LA-ICP-MS zircon U-Pb geochronology on the Helanshan ductile shear zones (HDSZ) in the Khondalite Belt. The results show that three pre-kinematic leucocratic dykes have been reworked by the shear zones and yielded magmatic zircon ages of 1934 ± 13 Ma, 1924 ± 5 Ma and 1920 ± 24 Ma. Two syn-kinematic leucocratic dykes appeared along the mylonitic foliations and were dated at 1887 ± 21 Ma and 1871 ± 28 Ma. A post-kinematic leucocratic dyke cut the mylonitic foliations and gave a crystallization age of 1816 ± 28 Ma. These emplacement ages of shear zone-related dykes constrain the development of the HDSZ at some time between ∼ 1920 Ma and ∼ 1816 Ma. Meanwhile, zircon overgrowth rims from three intrusions that suffered the shear deformation exhibited metamorphic ages of 1904 ± 11 Ma, 1840 ± 13 Ma and 1826 ± 31 Ma. Similar metamorphic ages of 1901 ± 7 Ma, 1888 ± 8 Ma and 1849 ± 29 Ma were also obtained from high-temperature felsic, pelitic and calcitic mylonites. These results further unravel that the HDSZ most likely developed at ca. 1904–1826 Ma. Additionally, we have summarized available deformation-related geochronological data in the Khondalite Belt, and proposed that this belt underwent three major phases of deformation (D1-D3) in the Orosirian. D1 mainly produced overturned to recumbent isoclinal folds F1, penetrative transposition foliations S1 and mineral lineations L1 at ca. 1.97–1.93 Ga. Subsequently, D2 dominantly resulted in tight to open upright folds F2 at ca. 1.93–1.90 Ga. Later, D3 generated a series of NE- to E-trending orogen-parallel ductile shear zones at ca. 1.90–1.82 Ga. Together with multiple high-grade metamorphism and magmatism, the D1-D3 deformation recorded a complex protracted orogeny (>100 Myr) in response to the collision between the Yinshan and Ordos blocks.

An enhanced quasi-3D HSDT for free vibration analysis of porous FG-CNT beams on a new concept of orthotropic VE-foundations

The original primary objective of this study is to conduct a comprehensive investigation into the free vibration behavior of beams with porosity, reinforced by carbon nanotubes (CNTs). These beams are supported by an arbitrary orthotropic variable elastic foundation (AOVEF). The focus lies on understanding how the presence of porosity and CNT reinforcement, coupled with the complex support provided by the AOVEF, influences the vibration characteristics of the beams. In addition, we intend to examine the impact of a number of micromechanical models on the vibration properties of these beams. Four carbon nanotube variables are introduced to symbolize several CNT variations. CNTs’ mechanical properties are examined through a variety of micromechanical models. Shear deformation effects are considered into account in the framework of higher-order shear deformation (HSDT) beam theory. The equations of motion are constructed using Hamilton’s concept and the equation system for the FG-CNT beam with supported ends is solved via the Navier methodology. A new approach in elastic foundation modeling is the Arbitrarily Orthotropic Variable Elastic (AOP-VE) foundation. It builds on the Winkler layer concept but introduces the idea of a variable response along the length of the beam foundation. Unlike the Winkler-Pasternak model, AOP-VE allows for control over the directional properties of the Pasternak foundation. The study looks into multiple aspects, involving various distribution kinds of CNTs, the impact of Winkler and Pasternak factors, mode values, side-to-length ratio, porosity, and angle variation. The results indicate that these parameters have a major effect on the inherent vibration features of FG-CNT beams. The study results in various novel findings that improve the understanding of the subject topic and set a standard for future studies.

A deep learning-based dynamic deformable adaptive framework for locating the root region of the dynamic flames.

Traditional optical flame detectors (OFDs) in flame detection are susceptible to environmental interference, which will inevitably cause detection errors and miscalculations when confronted with a complex environment. The conventional deep learning-based models can mitigate the interference of complex environments by flame image feature extraction, which significantly improves the precision of flame recognition. However, these models focus on identifying the general profile of the static flame, but neglect to effectively locate the source of the dynamic flame. Therefore, this paper proposes a novel dynamic flame detection method named Dynamic Deformable Adaptive Framework (DDAF) for locating the flame root region dynamically. Specifically, to address limitations in flame feature extraction of existing detection models, the Deformable Convolution Network v2 (DCNv2) is introduced for more flexible adaptation to the deformations and scale variations of target objects. The Context Augmentation Module (CAM) is used to convey flame features into Dynamic Head (DH) to feature extraction from different aspects. Subsequently, the Layer-Adaptive Magnitude-based Pruning (LAMP) where the connection with the smallest LAMP score is pruned sequentially is employed to further enhance the speed of model detection. More importantly, both the coarse- and fine-grained location techniques are designed in the Inductive Modeling (IM) to accurately delineate the flame root region for effective fire control. Additionally, the Temporal Consistency-based Detection (TCD) contributes to improving the robustness of model detection by leveraging the temporal information presented in consecutive frames of a video sequence. Compared with the classical deep learning method, the experimental results on the custom flame dataset demonstrate that the AP0.5 value is improved by 4.4%, while parameters and FLOPs are reduced by 25.3% and 25.9%, respectively. The framework of this research extends applicability to a variety of flame detection scenarios, including industrial safety and combustion process control.

Deep-learning-based joint rigid and deformable contour propagation for magnetic resonance imaging-guided prostate radiotherapy.

Deep learning-based unsupervised image registration has recently been proposed, promising fast registration. However, it has yet to be adopted in the online adaptive magnetic resonance imaging-guided radiotherapy (MRgRT)workflow. In this paper, we design an unsupervised, joint rigid, and deformable registration framework for contour propagation in MRgRT of prostatecancer. Three-dimensional pelvic T2-weighted MRIs of 143 prostate cancer patients undergoing radiotherapy were collected and divided into 110, 13, and 20 patients for training, validation, and testing. We designed a framework using convolutional neural networks (CNNs) for rigid and deformable registration. We selected the deformable registration network architecture among U-Net, MS-D Net, and LapIRN and optimized the training strategy (end-to-end vs. sequential). The framework was compared against an iterative baseline registration. We evaluated registration accuracy (the Dice and Hausdorff distance of the prostate and bladder contours), structural similarity index, and folding percentage to compare the methods. We also evaluated the framework's robustness to rigid and elastic deformations and bias fieldperturbations. The end-to-end trained framework comprising LapIRN for the deformable component achieved the best median (interquartile range) prostate and bladder Dice of0.89 (0.85-0.91) and 0.86 (0.80-0.91), respectively. This accuracy was comparable to the iterative baseline registration: prostate and bladder Dice of 0.91 (0.88-0.93) and 0.86 (0.80-0.92). The best models complete rigid and deformable registration in 0.002 (0.0005) and 0.74 (0.43) s (Nvidia Tesla V100-PCIe 32 GB GPU), respectively. We found that the models are robust to translations up to 52 mm, rotations up to 15 , elastic deformations up to 40 mm, and biasfields. Our proposed unsupervised, deep learning-based registration framework can perform rigid and deformable registration in less than a second with contour propagation accuracy comparable with iterative registration.

Deep-DM: Deep-driven deformable model for 3D image segmentation using limited data.

Objective - Medical image segmentation is essential for several clinical tasks, including diagnosis, surgical and treatment planning, and image-guided interventions. Deep Learning (DL) methods have become the state-of-the-art for several image segmentation scenarios. However, a large and well-annotated dataset is required to effectively train a DL model, which is usually difficult to obtain in clinical practice, especially for 3D images. Methods - In this paper, we proposed Deep-DM, a learning-guided deformable model framework for 3D medical imaging segmentation using limited training data. In the proposed method, an energy function is learned by a Convolutional Neural Network (CNN) and integrated into an explicit deformable model to drive the evolution of an initial surface towards the object to segment. Specifically, the learning-based energy function is iteratively retrieved from localized anatomical representations of the image containing the image information around the evolving surface at each iteration. By focusing on localized regions of interest, this representation excludes irrelevant image information, facilitating the learning process. Results and conclusion - The performance of the proposed method is demonstrated for the tasks of left ventricle and fetal head segmentation in ultrasound, left atrium segmentation in Magnetic Resonance, and bladder segmentation in Computed Tomography, using different numbers of training volumes in each study. The results obtained showed the feasibility of the proposed method to segment different anatomical structures in different imaging modalities. Moreover, the results also showed that the proposed approach is less dependent on the size of the training dataset in comparison with state-of-the-art DL-based segmentation methods, outperforming them for all tasks when a low number of samples is available. Significance - Overall, by offering a more robust and less data-intensive approach to accurately segmenting anatomical structures, the proposed method has the potential to enhance clinical tasks that require image segmentation strategies.

Quantitative analysis of microstructure evolution, stress partitioning and thermodynamics in the dynamic transformation of Fe-14Ni alloy

Dynamic transformation (DT) of austenite (γ) to ferrite (α) in the hot deformation of various carbon steels was widely investigated. However, the nature of DT remains unclear due to the lack of quantitative analysis of stress partitioning between two phases and the uncertainty of local distribution of substitutional elements at the interface in multi-component carbon steels used in the previous studies. Therefore, in the present study, a binary Fe-Ni alloy with α + γ duplex microstructure in equilibrium was prepared and isothermally compressed in α + γ two-phase region to achieve a quantitative analysis of microstructure evolution, stress partitioning, and thermodynamics during DT. γ to α DT during isothermal compression and α to γ reverse transformation on isothermal annealing under unloaded condition after deformation were accompanied by Ni partitioning. The lattice strains during thermomechanical processing were obtained via in-situ neutron diffraction measurement, based on which the stress partitioning behavior between γ and α was discussed by using the generalized Hooke's law. A thermodynamic framework for the isothermal deformation in solids was established based on the basic laws of thermodynamics, and it was shown that the total Helmholtz free energy change in the deformable material during the isothermal process should be smaller than the work done to the deformable material. Under the present thermodynamic framework, the microstructure evolution in the isothermal compression of Fe-14Ni alloy was well explained by considering the changes in chemical free energy, plastic and elastic energies, and the work done to the material. In addition, the stabilization of the soft α phase in Fe-14Ni alloy by deformation was rationalized since the γ to α transformation decreased the total Helmholtz free energy by decreasing the elastic and dislocation energies.