Digital synthesis of free-form multimaterial structures for realization of arbitrary programmed mechanical responses.

SignificanceCreating structures to realize function-oriented mechanical responses is desired for many applications. Yet, the use of a single material phase and heuristics-based designs may fail to attain specific target behaviors. Here, through a deterministic algorithmic procedure, multiple materials with dissimilar properties are intelligently synthesized into composite structures to achieve arbitrary prescribed responses. Created structures possess unconventional geometry and seamless integration of multiple materials. Despite geometric complexity and varied material phases, these structures are fabricated by multimaterial manufacturing, and tested to demonstrate that wide-ranging nonlinear responses are physically and accurately realized. Upon heteroassembly, resulting structures provide architectures that exhibit highly complex yet navigable responses. The proposed strategy can benefit the design of function-oriented nonlinear mechanical devices, such as actuators and energy absorbers.

Eulerian simulation of complex suspensions and biolocomotion in three dimensions.

We present a numerical method specifically designed for simulating three-dimensional fluid-structure interaction (FSI) problems based on the reference map technique (RMT). The RMT is a fully Eulerian FSI numerical method that allows fluids and large-deformation elastic solids to be represented on a single fixed computational grid. This eliminates the need for meshing complex geometries typical in other FSI approaches and greatly simplifies the coupling between fluid and solids. We develop a three-dimensional implementation of the RMT, parallelized using the distributed memory paradigm, to simulate incompressible FSI with neo-Hookean solids. As part of our method, we develop a field extrapolation scheme that works efficiently in parallel. Through representative examples, we demonstrate the method's suitability in investigating many-body and active systems, as well as its accuracy and convergence. The examples include settling of a mixture of heavy and buoyant soft ellipsoids, lid-driven cavity flow containing a soft sphere, and swimmers actuated via active stress.

Effects of interfacial debonding on the stability of finitely strained periodic microstructured elastic composites.

Predicting failure initiation in nonlinear composite materials, often referred to as metamaterials, is a fundamental challenge in nonlinear solid mechanics. Microstructural failure mechanisms encompass fracture, decohesion, cavitation, compression-induced contact and instabilities, affecting their unconventional static and dynamic performances. To fully take advantage of these materials, especially in extreme applications, it is imperative to predict their nonlinear behaviour using reliable, accurate and computationally efficient numerical methodologies. This study presents an innovative nonlinear homogenization-based theoretical framework for characterizing the failure behaviour of periodic reinforced hyperelastic composites induced by reinforcement/matrix decohesion and interaction between contact mechanisms and microscopic instabilities. Debonding and unilateral contact between different phases are incorporated by employing an enhanced cohesive/contact model, which features a special nonlinear interface constitutive law and an accurate contact formulation within the context of finite strain continuum mechanics. The theoretical formulation is demonstrated using periodically layered composites subjected to macroscopic compressive loading conditions along the lamination direction. Numerical results illustrate the ways in which debonding phenomena, in conjunction with fibre microbuckling, may influence the critical loads of the examined composite solid. The sensitivity of the results obtained through the proposed contact-cohesive model at finite strain with respect to its implementation is also explored. This article is part of the theme issue 'Current developments in elastic and acoustic metamaterials science (Part 1)'.

Decoupling Transmission and Transduction for Improved Durability of Highly Stretchable, Soft Strain Sensing: Applications in Human Health Monitoring.

This work presents a modular approach to the development of strain sensors for large deformations. The proposed method separates the extension and signal transduction mechanisms using a soft, elastomeric transmission and a high-sensitivity microelectromechanical system (MEMS) transducer. By separating the transmission and transduction, they can be optimized independently for application-specific mechanical and electrical performance. This work investigates the potential of this approach for human health monitoring as an implantable cardiac strain sensor for measuring global longitudinal strain (GLS). The durability of the sensor was evaluated by conducting cyclic loading tests over one million cycles, and the results showed negligible drift. To account for hysteresis and frequency-dependent effects, a lumped-parameter model was developed to represent the viscoelastic behavior of the sensor. Multiple model orders were considered and compared using validation and test data sets that mimic physiologically relevant dynamics. Results support the choice of a second-order model, which reduces error by 73% compared to a linear calibration. In addition, we evaluated the suitability of this sensor for the proposed application by demonstrating its ability to operate on compliant, curved surfaces. The effects of friction and boundary conditions are also empirically assessed and discussed.

Formation and size distribution of self-assembled vesicles.

When detergents and phospholipid membranes are dispersed in aqueous solutions, they tend to self-assemble into vesicles of various shapes and sizes by virtue of their hydrophobic and hydrophilic segments. A clearer understanding of such vesiculation processes holds promise for better elucidation of human physiology and disease, and paves the way to improved diagnostics, drug development, and drug delivery. Here we present a detailed analysis of the energetics and thermodynamics of vesiculation by recourse to nonlinear elasticity, taking into account large deformation that may arise during the vesiculation process. The effects of membrane size, spontaneous curvature, and membrane stiffness on vesiculation and vesicle size distribution were investigated, and the critical size for vesicle formation was determined and found to compare favorably with available experimental evidence. Our analysis also showed that the critical membrane size for spontaneous vesiculation was correlated with membrane thickness, and further illustrated how the combined effects of membrane thickness and physical properties influenced the size, shape, and distribution of vesicles. These findings shed light on the formation of physiological extracellular vesicles, such as exosomes. The findings also suggest pathways for manipulating the size, shape, distribution, and physical properties of synthetic vesicles, with potential applications in vesicle physiology, the pathobiology of cancer and other diseases, diagnostics using in vivo liquid biopsy, and drug delivery methods.

Semantically Guided Large Deformation Estimation with Deep Networks.

Deformable image registration is still a challenge when the considered images have strong variations in appearance and large initial misalignment. A huge performance gap currently remains for fast-moving regions in videos or strong deformations of natural objects. We present a new semantically guided and two-step deep deformation network that is particularly well suited for the estimation of large deformations. We combine a U-Net architecture that is weakly supervised with segmentation information to extract semantically meaningful features with multiple stages of nonrigid spatial transformer networks parameterized with low-dimensional B-spline deformations. Combining alignment loss and semantic loss functions together with a regularization penalty to obtain smooth and plausible deformations, we achieve superior results in terms of alignment quality compared to previous approaches that have only considered a label-driven alignment loss. Our network model advances the state of the art for inter-subject face part alignment and motion tracking in medical cardiac magnetic resonance imaging (MRI) sequences in comparison to the FlowNet and Label-Reg, two recent deep-learning registration frameworks. The models are compact, very fast in inference, and demonstrate clear potential for a variety of challenging tracking and/or alignment tasks in computer vision and medical image analysis.

An efficient and accurate method for modeling nonlinear fractional viscoelastic biomaterials.

Computational biomechanics plays an important role in biomedical engineering: using modeling to understand pathophysiology, treatment and device design. While experimental evidence indicates that the mechanical response of most tissues is viscoelastic, current biomechanical models in the computational community often assume hyperelastic material models. Fractional viscoelastic constitutive models have been successfully used in literature to capture viscoelastic material response; however, the translation of these models into computational platforms remains limited. Many experimentally derived viscoelastic constitutive models are not suitable for three-dimensional simulations. Furthermore, the use of fractional derivatives can be computationally prohibitive, with a number of current numerical approximations having a computational cost that is 𝒪 ( N T 2 ) and a storage cost that is 𝒪(NT ) (NT denotes the number of time steps). In this paper, we present a novel numerical approximation to the Caputo derivative which exploits a recurrence relation similar to those used to discretize classic temporal derivatives, giving a computational cost that is 𝒪(NT ) and a storage cost that is fixed over time. The approximation is optimized for numerical applications, and an error estimate is presented to demonstrate the efficacy of the method. The method, integrated into a finite element solid mechanics framework, is shown to be unconditionally stable in the linear viscoelastic case. It was then integrated into a computational biomechanical framework, with several numerical examples verifying the accuracy and computational efficiency of the method, including in an analytic test, in an analytic fractional differential equation, as well as in a computational biomechanical model problem.

Measurement of large strain properties in calf muscles in vivo using magnetic resonance elastography and spatial modulation of magnetization.

It is important to measure the large deformation properties of skeletal muscle in vivo in order to understand and model movement and the force-producing capabilities of muscle. As muscle properties are non-linear, an understanding of how the deformation state affects the measured shear moduli is also useful for clinical applications of magnetic resonance elastography (MRE) to muscle disorders. MRE has so far only been used to measure the linear viscoelastic (small strain) properties of muscles. This study aims to measure the shear moduli of human calf muscles under varying degrees of strain using MRE. Nine healthy adults (four males; age range, 25-38 years) were recruited, and the storage modulus G' was measured at three ankle angle positions: P0 (neutral), P15 (15° plantarflexed) and P30 (30° plantarflexed). Spatial modulation of magnetization (SPAMM) was used to measure the strain in the calf associated with the ankle rotations between P0 to P15 and P0 to P30. SPAMM results showed that, with plantarflexion, there was a shortening of the medial gastrocnemius and soleus muscles, which resulted in an expansion of both muscles in the transverse direction. Strains for each ankle rotation were in the range 3-9% (in compression). MRE results showed that this shortening during plantarflexion resulted in a mean decrease in G' in the medial gastrocnemius (p = 0.013, linear mixed model), but not in the soleus (p = 0.47). This study showed that MRE is a viable technique for the measurement of large strain deformation properties in vivo in soft tissues by inducing physiological strain within the muscle during imaging.

Finite element analysis of biological soft tissue surrounded by a deformable membrane that controls transmembrane flow.

BACKGROUND: Many biological soft tissues are hydrated porous hyperelastic materials, which consist of a complex solid skeleton with fine voids and fluid filling these voids. Mechanical interactions between the solid and the fluid in hydrated porous tissues have been analyzed by finite element methods (FEMs) in which the mixture theory was introduced in various ways. Although most of the tissues are surrounded by deformable membranes that control transmembrane flows, the boundaries of the tissues have been treated as rigid and/or freely permeable in these studies. The purpose of this study was to develop a method for the analysis of hydrated porous hyperelastic tissues surrounded by deformable membranes that control transmembrane flows. RESULTS: For this, we developed a new nonlinear finite element formulation of the mixture theory, where the nodal unknowns were the pore water pressure and solid displacement. This method allows the control of the fluid flow rate across the membrane using Neumann boundary condition. Using the method, we conducted a compression test of the hydrated porous hyperelastic tissue, which was surrounded by a flaccid impermeable membrane, and a part of the top surface of this tissue was pushed by a platen. The simulation results showed a stress relaxation phenomenon, resulting from the interaction between the elastic deformation of the tissue, pore water pressure gradient, and the movement of fluid. The results also showed that the fluid trapped by the impermeable membrane led to the swelling of the tissue around the platen. CONCLUSIONS: These facts suggest that our new method can be effectively used for the analysis of a large deformation of hydrated porous hyperelastic material surrounded by a deformable membrane that controls transmembrane flow, and further investigations may allow more realistic analyses of the biological soft tissues, such as brain edema, brain trauma, the flow of blood and lymph in capillaries and pitting edema.

The same modality medical image registration with large deformation and clinical application based on adaptive diffeomorphic multi-resolution demons.

BACKGROUND: Diffeomorphic demons can not only guarantee smooth and reversible deformation, but also avoid unreasonable deformation. However, the number of iterations which has great influence on the registration result needs to be set manually. METHODS: This study proposed a novel method to exploit the adaptive diffeomorphic multi-resolution demons algorithm to the non-rigid registration of the same modality medical images with large deformation. Firstly an optimized non-rigid registration framework and the diffeomorphism strategy were used, and then a similarity energy function based on the grey value was designed as registration metric, lastly termination condition was set based on the variation of this metric and iterations can be stopped adaptively. Quantitative analyses based on the registration evaluation indexes were conducted to prove the validity of this method. RESULTS: Registration result of synthetic image and the same modality MRI and CT image was compared with those obtained by other demons algorithms. Quantitative analyses demonstrated the proposed method's superiority. Medical image with large deformation was produced by rotational distortion and extrusion transform, and the same modality image registration with large deformation was performed successfully. Quantitative analyses showed that the registration evaluation indexes remained stable with an increase in transform strength. This method can be also applied to pulmonary medical image registration with large deformation successfully, and it showed the clinical application value. The influence of different driving forces and parameters on the registration result was analysed, and the result demonstrated that the proposed method is effective and robust. CONCLUSIONS: This method can solve the non-rigid registration problem of the same modality medical image with large deformation showing promise for diagnostic pulmonary imaging applications.

Combining tract- and atlas-based analysis reveals microstructural abnormalities in early Tourette syndrome children.

Tourette syndrome (TS) is a neurological disorder that causes uncontrolled repetitive motor and vocal tics in children. Examining the neural basis of TS churned out different research studies that advanced our understanding of the brain pathways involved in its development. Particularly, growing evidence points to abnormalities within the fronto-striato-thalamic pathways. In this study, we combined Tract-Based Spatial Statistics (TBSS) and Atlas-based regions of interest (ROI) analysis approach, to investigate the microstructural diffusion changes in both deep and superficial white matter (SWM) in TS children. We then characterized the altered microstructure of white matter in 27 TS children in comparison with 27 age- and gender-matched healthy controls. We found that fractional anisotropy (FA) decreases and radial diffusivity (RD) increases in deep white matter (DWM) tracts in cortico-striato-thalamo-cortical (CSTC) circuit as well as SWM. Furthermore, we found that lower FA values and higher RD values in white matter regions are correlated with more severe tics, but not tics duration. Besides, we also found both axial diffusivity and mean diffusivity increase using Atlas-based ROI analysis. Our work may suggest that microstructural diffusion changes in white matter is not only restricted to the gray matter of CSTC circuit but also affects SWM within the primary motor and somatosensory cortex, commissural and association fibers. Hum Brain Mapp 37:1903-1919, 2016. © 2016 Wiley Periodicals, Inc.

Automatic whole brain tract-based analysis using predefined tracts in a diffusion spectrum imaging template and an accurate registration strategy.

Automated tract-based analysis of diffusion MRI is an important tool for investigating tract integrity of the cerebral white matter. Current template-based automatic analyses still lack a comprehensive list of tract atlas and an accurate registration method. In this study, tract-based automatic analysis (TBAA) was developed to meet the demands. Seventy-six major white matter tracts were reconstructed on a high-quality diffusion spectrum imaging (DSI) template, and an advanced two-step registration strategy was proposed by incorporating anatomical information of the gray matter from T1-weighted images in addition to microstructural information of the white matter from diffusion-weighted images. The automatic analysis was achieved by establishing a transformation between the DSI template and DSI dataset of the subject derived from the registration strategy. The tract coordinates in the template were transformed to native space in the individual's DSI dataset, and the microstructural properties of major tract bundles were sampled stepwise along the tract coordinates of the subject's DSI dataset. In a validation study of eight well-known tracts, our results showed that TBAA had high geometric agreement with manual tracts in both deep and superficial parts but significantly smaller measurement variability than manual method in functional difference. Additionally, the feasibility of the method was demonstrated by showing tracts with altered microstructural properties in patients with schizophrenia. Fifteen major tract bundles were found to have significant differences after controlling the family-wise error rate. In conclusion, the proposed TBAA method is potentially useful in brain-wise investigations of white matter tracts, particularly for a large cohort study.

Shape abnormalities of subcortical and ventricular structures in mild cognitive impairment and Alzheimer's disease: detecting, quantifying, and predicting.

This article assesses the feasibility of using shape information to detect and quantify the subcortical and ventricular structural changes in mild cognitive impairment (MCI) and Alzheimer's disease (AD) patients. We first demonstrate structural shape abnormalities in MCI and AD as compared with healthy controls (HC). Exploring the development to AD, we then divide the MCI participants into two subgroups based on longitudinal clinical information: (1) MCI patients who remained stable; (2) MCI patients who converted to AD over time. We focus on seven structures (amygdala, hippocampus, thalamus, caudate, putamen, globus pallidus, and lateral ventricles) in 754 MR scans (210 HC, 369 MCI of which 151 converted to AD over time, and 175 AD). The hippocampus and amygdala were further subsegmented based on high field 0.8 mm isotropic 7.0T scans for finer exploration. For MCI and AD, prominent ventricular expansions were detected and we found that these patients had strongest hippocampal atrophy occurring at CA1 and strongest amygdala atrophy at the basolateral complex. Mild atrophy in basal ganglia structures was also detected in MCI and AD. Stronger atrophy in the amygdala and hippocampus, and greater expansion in ventricles was observed in MCI converters, relative to those MCI who remained stable. Furthermore, we performed principal component analysis on a linear shape space of each structure. A subsequent linear discriminant analysis on the principal component values of hippocampus, amygdala, and ventricle leads to correct classification of 88% HC subjects and 86% AD subjects.

Research on stress field inversion and large deformation level determination of super deep buried soft rock tunnel.

Understanding the characteristics and distribution patterns of the initial geo-stress field in tunnels is of great significance for studying the problem of large deformation of tunnels under high geo-stress conditions. This article proposes a ground stress field inversion method and large deformation level determination based on the GS-XGBoost algorithm and the Haba Snow Mountain Tunnel of the Lixiang Railway. Firstly, the hydraulic fracturing method is used to conduct on-site testing of tunnel ground stress and obtain tunnel ground stress data. Then, a three-dimensional model of the Haba Snow Mountain Tunnel will be established, and it will be combined with the GS-XGBoost regression algorithm model to obtain the optimal boundary conditions of the model. Finally, the optimal boundary condition parameters are substituted into the three-dimensional finite-difference calculation model for stress calculation, and the distribution of the in-situ stress field of the entire calculation model is obtained. Finally, the level of large deformation of the Haba Snow Mountain Tunnel will be determined. The results show that the ground stress of the tunnel increases with the increase of burial depth, with the maximum horizontal principal stress of 38.03 MPa and the minimum horizontal principal stress of 26.07 MPa. The Haba Snow Mountain Tunnel has large deformation problems of levels I, II, III, and IV. Level III and IV large deformations are generally accompanied by higher ground stress (above 28 MPa) and smaller surrounding rock strength. The distribution of surrounding rock strength along the tunnel axis shows a clear "W" shape, opposite to the surface elevation "M" shape. It is inferred that the mountain may be affected by geological structures on both sides of the north and south, causing more severe compression of the tunnel surrounding rock at the peak.

Analytical and numerical solutions of pore formation in elastic food materials during dehydration.

In this paper, we describe a model for pore formation in food materials during drying. As a proxy for fruits and vegetables, we take a spherical hydrogel, with a stiff elastic skin, and a central cavity filled with air and water vapour. The model describes moisture transport coupled to large deformation mechanics. Both stress and chemical potential are derived from a free energy functional, following the framework developed by Suo and coworkers. We have compared Finite Volume and Finite Element implementations and analytical solutions with each other, and we show that they render similar solutions. The Finite Element solver has a larger range of numerical stability than the Finite Volume solver, and the analytical solution also has a limited range of validity. Since the Finite Element solver operates using the mathematically intricate weak form, we introduce the method in a tutorial manner for food scientists. Subsequently, we have explored the physics of the pore formation problem further with the Finite Element solver. We show that the presence of an elastic skin is a prerequisite for the growth of the central cavity. The elastic skin must have an elastic modulus of at least 10 times that of the hydrogel. An initial pore with 10% of the size of the gel can grow to 5 times its initial size. Such an increase in porosity has been reported in the literature on drying of vegetables, if a dense hard skin is formed, known as case hardening. We discuss that models as presented in this paper, where moisture transport is strongly coupled to large deformation mechanics, are required if one wants to describe pore/structure formation during drying and intensive heating (as baking and frying) of food materials from first principles.

QUIZ: An arbitrary volumetric point matching method for medical image registration.

Rigid pre-registration involving local-global matching or other large deformation scenarios is crucial. Current popular methods rely on unsupervised learning based on grayscale similarity, but under circumstances where different poses lead to varying tissue structures, or where image quality is poor, these methods tend to exhibit instability and inaccuracies. In this study, we propose a novel method for medical image registration based on arbitrary voxel point of interest matching, called query point quizzer (QUIZ). QUIZ focuses on the correspondence between local-global matching points, specifically employing CNN for feature extraction and utilizing the Transformer architecture for global point matching queries, followed by applying average displacement for local image rigid transformation.We have validated this approach on a large deformation dataset of cervical cancer patients, with results indicating substantially smaller deviations compared to state-of-the-art methods. Remarkably, even for cross-modality subjects, it achieves results surpassing the current state-of-the-art.

Independently Trained Multi-Scale Registration Network Based on Image Pyramid.

Image registration is a fundamental task in various applications of medical image analysis and plays a crucial role in auxiliary diagnosis, treatment, and surgical navigation. However, cardiac image registration is challenging due to the large non-rigid deformation of the heart and the complex anatomical structure. To address this challenge, this paper proposes an independently trained multi-scale registration network based on an image pyramid. By down-sampling the original input image multiple times, we can construct image pyramid pairs, and design a multi-scale registration network using image pyramid pairs of different resolutions as the training set. Using image pairs of different resolutions, train each registration network independently to extract image features from the image pairs at different resolutions. During the testing stage, the large deformation registration is decomposed into a multi-scale registration process. The deformation fields of different resolutions are fused by a step-by-step deformation method, thereby addressing the challenge of directly handling large deformations. Experiments were conducted on the open cardiac dataset ACDC (Automated Cardiac Diagnosis Challenge); the proposed method achieved an average Dice score of 0.828 in the experimental results. Through comparative experiments, it has been demonstrated that the proposed method effectively addressed the challenge of heart image registration and achieved superior registration results for cardiac images.

Understanding bacterial infiltration of the pancreas through a deformable pancreatic duct.

Tiny amount of bacteria are found in the pancreas in pancreatitis and cancer, which seemed involved in inflammation and carcinogenesis. However, bacterial infiltration from the duodenum is inhibited by the physical defense mechanisms such as bile flow and the sphincter of Oddi. To understand how the bacteria possibly infiltrate the pancreas through a deformable pancreatic duct, influenced by the periodic contractions of the sphincter of Oddi, a mathematical model of bacterial infiltration is developed that considered large deformation, fluid flow, and bacterial transport in a deformable pancreatic duct. In addition, the sphincter's contraction wave is modeled by including its propagation from the pancreas toward the duodenum. Simulated structure of the deformed duct with the relaxed sphincter and simulated bile distribution agreed reasonably well with the literature, validating the model. Bacterial infiltration from the duodenum in a deformable pancreatic duct, following the sphincter's contraction, is counteracted by a gradual peristalsis-like deformation of the pancreatic duct, due to an antegrade contraction wave propagation from the pancreas to the duodenum, Parametric sensitivity analysis demonstrated that bacterial infiltration is increased with lower bile and pancreatic juice flow rate, greater contraction amplitude and frequency, thinner wall thickness, and retrograde contraction wave propagation. Since contraction waves following retrograde propagation are increased in patients with common bile duct stones and pancreatitis, they may possibly be factors for continuum inflammation of pancreas. (224 words).

A systematic comparison between FEBio and PolyFEM for biomechanical systems.

BACKGROUND AND OBJECTIVES: Finite element simulations are widely employed as a non-invasive and cost-effective approach for predicting outcomes in biomechanical simulations. However, traditional finite element software, primarily designed for engineering materials, often encountered limitations in contact detection and enforcement, leading to simulation failure when dealing with complex biomechanical configurations. Currently, a lot of model tuning is required to get physically accurate finite element simulations without failures. This adds significant human interaction to each iteration of a biomechanical model. This study addressed these issues by introducing PolyFEM, a novel finite element solver that guarantees inversion- and intersection-free solutions with completely automatic collision detection. The objective of this research is to validate PolyFEM's capabilities by comparing its results with those obtained from a well-established finite element solver, FEBio. METHODS: To achieve this goal, five comparison scenarios were formulated to assess and validate PolyFEM's performance. The simulations were reproduced using both PolyFEM and FEBio, and the final results were compared. The five comparison scenarios included: (1) reproducing simulations from the FEBio test suite, consisting of static, dynamic, and contact-driven simulations; (2) replicating simulations from the verification paper published alongside the original release of FEBio; (3) a biomechanically based contact problem; (4) creating a custom simulation involving high-energy collisions between soft materials to highlight the difference in collision methods between the two solvers; and (5) performing biomechanical simulations of biting and quasi-stance. RESULTS: We found that PolyFEM was capable of replicating all simulations previously conducted in FEBio. Particularly noteworthy is PolyFEM's superiority in high-energy contact simulations, where FEBio fell short, unable to complete over half of the simulations in Scenario 4. Although some of the simulations required significantly more simulation time in PolyFEM compared to FEBio, it is important to highlight that PolyFEM achieved these results without the need for any additional model tuning or contact declaration. DISCUSSION: Despite being in the early stages of development, PolyFEM currently provides verified solutions for hyperelastic materials that are consistent with FEBio, both in previously published workflows and novel finite element scenarios. PolyFEM exhibited the ability to tackle challenging biomechanical problems where other solvers fell short, thus offering the potential to enhance the accuracy and realism of future finite element analyses.

Unsupervised lung CT image registration via stochastic decomposition of deformation fields.

We address the problem of lung CT image registration, which underpins various diagnoses and treatments for lung diseases. The main crux of the problem is the large deformation that the lungs undergo during respiration. This physiological process imposes several challenges from a learning point of view. In this paper, we propose a novel training scheme, called stochastic decomposition, which enables deep networks to effectively learn such a difficult deformation field during lung CT image registration. The key idea is to stochastically decompose the deformation field, and supervise the registration by synthetic data that have the corresponding appearance discrepancy. The stochastic decomposition allows for revealing all possible decompositions of the deformation field. At the learning level, these decompositions can be seen as a prior to reduce the ill-posedness of the registration yielding to boost the performance. We demonstrate the effectiveness of our framework on Lung CT data. We show, through extensive numerical and visual results, that our technique outperforms existing methods.