This study aims to establish an artificial intelligence model, ThyroidNet, to diagnose thyroid nodules using deep learning techniques accurately.

The topological connectivity information derived from the brain functional network can bring new insights for diagnosing and analyzing dementia disorders. The brain functional network is suitable to bridge the correlation between abnormal connectivities and dementia disorders. However, it is challenging to access considerable amounts of brain functional network data, which hinders the widespread application of data-driven models in dementia diagnosis. In this study, a novel distribution-regularized adversarial graph auto-Encoder (DAGAE) with transformer is proposed to generate new fake brain functional networks to augment the brain functional network dataset, improving the dementia diagnosis accuracy of data-driven models. Specifically, the label distribution is estimated to regularize the latent space learned by the graph encoder, which can make the learning process stable and the learned representation robust. Also, the transformer generator is devised to map the node representations into node-to-node connections by exploring the long-term dependence of highly-correlated distant brain regions. The typical topological properties and discriminative features can be preserved entirely. Furthermore, the generated brain functional networks improve the prediction performance using different classifiers, which can be applied to analyze other cognitive diseases. Attempts on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset demonstrate that the proposed model can generate good brain functional networks. The classification results show adding generated data can achieve the best accuracy value of 85.33%, sensitivity value of 84.00%, specificity value of 86.67%. The proposed model also achieves superior performance compared with other related augmented models. Overall, the proposed model effectively improves cognitive disease diagnosis by generating diverse brain functional networks.

Over the years, the continuous development of new technology has promoted research in the field of posture recognition and also made the application field of posture recognition have been greatly expanded. The purpose of this paper is to introduce the latest methods of posture recognition and review the various techniques and algorithms of posture recognition in recent years, such as scale-invariant feature transform, histogram of oriented gradients, support vector machine (SVM), Gaussian mixture model, dynamic time warping, hidden Markov model (HMM), lightweight network, convolutional neural network (CNN). We also investigate improved methods of CNN, such as stacked hourglass networks, multi-stage pose estimation networks, convolutional pose machines, and high-resolution nets. The general process and datasets of posture recognition are analyzed and summarized, and several improved CNN methods and three main recognition techniques are compared. In addition, the applications of advanced neural networks in posture recognition, such as transfer learning, ensemble learning, graph neural networks, and explainable deep neural networks, are introduced. It was found that CNN has achieved great success in posture recognition and is favored by researchers. Still, a more in-depth research is needed in feature extraction, information fusion, and other aspects. Among classification methods, HMM and SVM are the most widely used, and lightweight network gradually attracts the attention of researchers. In addition, due to the lack of 3D benchmark data sets, data generation is a critical research direction.

For people all over the world, cancer is one of the most feared diseases. Cancer is one of the major obstacles to improving life expectancy in countries around the world and one of the biggest causes of death before the age of 70 in 112 countries. Among all kinds of cancers, breast cancer is the most common cancer for women. The data showed that female breast cancer had become one of the most common cancers.

It has been recognized that fluid-structure interactions (FSI) play an important role in cardiovascular disease initiation and development. However, in vivo MRI multi-component FSI models for human carotid atherosclerotic plaques with bifurcation and quantitative comparisons of FSI models with fluid-only or structure-only models are currently lacking in the literature. A 3D non-Newtonian multi-component FSI model based on in vivo/ex vivo MRI images for human atherosclerotic plaques was introduced to investigate flow and plaque stress/strain behaviors which may be related to plaque progression and rupture. Both artery wall and plaque components were assumed to be hyperelastic, isotropic, incompressible and homogeneous. Blood flow was assumed to be laminar, non-Newtonian, viscous and incompressible. In vivo/ex vivo MRI images were acquired using histologically-validated multi-spectral MRI protocols. The 3D FSI models were solved and results were compared with those from a Newtonian FSI model and wall-only/fluid-only models. A 145% difference in maximum principal stresses (Stress-P(1)) between the FSI and wall-only models and 40% difference in flow maximum shear stress (MSS) between the FSI and fluid-only models were found at the throat of the plaque using a severe plaque sample (70% severity by diameter). Flow maximum shear stress (MSS) from the rigid wall model is much higher (20-40% in maximum MSS values, 100-150% in stagnation region) than those from FSI models.

Atherosclerotic plaque rupture and progression have been the focus of intensive investigations in recent years. Plaque rupture is closely related to most severe cardiovascular syndromes such as heart attack and stroke. A computational procedure based on meshless generalized finite difference (MGFD) method and serial magnetic resonance imaging (MRI) data was introduced to quantify patient-specific carotid atherosclerotic plaque growth functions and simulate plaque progression. Participating patients were scanned three times (T(1), T(2), and T(3), at intervals of about 18 months) to obtain plaque progression data. Vessel wall thickness (WT) changes were used as the measure for plaque progression. Since there was insufficient data with the current technology to quantify individual plaque component growth, the whole plaque was assumed to be uniform, homogeneous, hyperelastic, isotropic and nearly incompressible. The linear elastic model was used. The 2D plaque model was discretized and solved using a meshless generalized finite difference (GFD) method. Starting from the T(2) plaque geometry, plaque progression was simulated by solving the solid model and adjusting wall thickness using plaque growth functions iteratively until T(3) is reached. Numerically simulated plaque progression agreed very well with actual plaque geometry at T(3) given by MRI data. We believe this is the first time plaque progression simulation based on multi-year patient-tracking data was reported. Serial MRI-based progression simulation adds time dimension to plaque vulnerability assessment and will improve prediction accuracy for potential plaque rupture risk.

We present a new method for simulating two-phase flows in complex geometries, taking into account contact lines separating immiscible incompressible components. We combine the diffuse domain method for solving PDEs in complex geometries with the diffuse-interface (phase-field) method for simulating multiphase flows. In this approach, the complex geometry is described implicitly by introducing a new phase-field variable, which is a smooth approximation of the characteristic function of the complex domain. The fluid and component concentration equations are reformulated and solved in larger regular domain with the boundary conditions being implicitly modeled using source terms. The method is straightforward to implement using standard software packages; we use adaptive finite elements here. We present numerical examples demonstrating the effectiveness of the algorithm. We simulate multiphase flow in a driven cavity on an extended domain and find very good agreement with results obtained by solving the equations and boundary conditions in the original domain. We then consider successively more complex geometries and simulate a droplet sliding down a rippled ramp in 2D and 3D, a droplet flowing through a Y-junction in a microfluidic network and finally chaotic mixing in a droplet flowing through a winding, serpentine channel. The latter example actually incorporates two different diffuse domains: one describes the evolving droplet where mixing occurs while the other describes the channel.

Cardiovascular disease (CVD) is becoming the number one cause of death worldwide. Atherosclerotic plaque rupture and progression are closely related to most severe cardiovascular syndromes such as heart attack and stroke. Mechanisms governing plaque rupture and progression are not well understood. A computational procedure based on three-dimensional meshless generalized finite difference (MGFD) method and serial magnetic resonance imaging (MRI) data was introduced to quantify patient-specific carotid atherosclerotic plaque growth functions and simulate plaque progression. Participating patients were scanned three times (T1, T2, and T3, at intervals of about 18 months) to obtain plaque progression data. Vessel wall thickness (WT) changes were used as the measure for plaque progression. Since there was insufficient data with the current technology to quantify individual plaque component growth, the whole plaque was assumed to be uniform, homogeneous, isotropic, linear, and nearly incompressible. The linear elastic model was used. The 3D plaque model was discretized and solved using a meshless generalized finite difference (GFD) method. Four growth functions with different combinations of wall thickness, stress, and neighboring point terms were introduced to predict future plaque growth based on previous time point data. Starting from the T2 plaque geometry, plaque progression was simulated by solving the solid model and adjusting wall thickness using plaque growth functions iteratively until T3 is reached. Numerically simulated plaque progression agreed very well with the target T3 plaque geometry with errors ranging from 11.56%, 6.39%, 8.24%, to 4.45%, given by the four growth functions. We believe this is the first time 3D plaque progression simulation based on multi-year patient-tracking data was reported. Serial MRI-based progression simulation adds time dimension to plaque vulnerability assessment and will improve prediction accuracy for potential plaque rupture risk.

Right and left ventricle (RV/LV) combination models with three different patch materials (Dacron scaffold, treated pericardium, and contracting myocardium), two-layer construction, fiber orientation, and active anisotropic material properties were introduced to evaluate the effects of patch materials on RV function. A material-stiffening approach was used to model active heart contraction. Cardiac magnetic resonance (CMR) imaging was performed to acquire patient-specific ventricular geometries and cardiac motion from a patient with severe RV dilatation due to pulmonary regurgitation needing RV remodeling and pulmonary valve replacement operation. Computational models were constructed and solved to obtain RV stroke volume, ejection fraction, patch area variations, and stress/strain data for patch comparisons. Our results indicate that the patch model with contracting myocardium leads to decreased stress level in the patch area, improved RV function and patch area contractility. Maximum Stress-P(1) (maximum principal stress) value at the center of the patch from the Dacron scaffold patch model was 350% higher than that from the other two models. Patch area reduction ratio was 0.3%, 3.1% and 27.4% for Dacron scaffold, pericardium, and contracting myocardium patches, respectively. These findings suggest that the contracting myocardium patch model may lead to improved recovery of RV function in patients with severe chronic pulmonary regurgitation.

Previously, we introduced a computational procedure based on three-dimensional meshless generalized finite difference (MGFD) method and serial magnetic resonance imaging (MRI) data to quantify patient-specific carotid atherosclerotic plaque growth functions and simulate plaque progression. Structure-only models were used in our previous report. In this paper, fluid-stricture interaction (FSI) was added to improve on prediction accuracy. One participating patient was scanned three times (T1, T2, and T3, at intervals of about 18 months) to obtain plaque progression data. Blood flow was assumed to laminar, Newtonian, viscous and incompressible. The Navier-Stokes equations with arbitrary Lagrangian-Eulerian (ALE) formulation were used as the governing equations. Plaque material was assumed to be uniform, homogeneous, isotropic, linear, and nearly incompressible. The linear elastic model was used. The 3D FSI plaque model was discretized and solved using a meshless generalized finite difference (GFD) method. Growth functions with a) morphology alone; b) morphology and plaque wall stress (PWS); morphology and flow shear stress (FSS), and d) morphology, PWS and FSS were introduced to predict future plaque growth based on previous time point data. Starting from the T2 plaque geometry, plaque progression was simulated by solving the FSI model and adjusting plaque geometry using plaque growth functions iteratively until T3 is reached. Numerically simulated plaque progression agreed very well with the target T3 plaque geometry with errors ranging from 8.62%, 7.22%, 5.77% and 4.39%, with the growth function including morphology, plaque wall stress and flow shear stress terms giving the best predictions. Adding flow shear stress term to the growth function improved the prediction error from 7.22% to 4.39%, a 40% improvement. We believe this is the first time 3D plaque progression FSI simulation based on multi-year patient-tracking data was reported. Serial MRI-based progression simulation adds time dimension to plaque vulnerability assessment and will improve prediction accuracy for potential plaque rupture risk.

Identifying ventricle material properties and its infarct area after heart attack noninvasively is of great important in clinical applications. An echo-based computational modeling approach was proposed to investigate left ventricle (LV) mechanical properties and stress conditions using patient-specific data. Echo data was acquired from one healthy volunteer (male, age: 58) and a male patient (age: 60) who had an acute inferior myocardial infarction one week before echo image acquisition. Standard echocardiograms were obtained using an ultrasound machine (E9, GE Mechanical Systems, Milwaukee, Wisconsin) with a 3V probe and data were segmented for model construction. Finite element models were constructed to obtain ventricle stress and strain conditions. A pre-shrink process was applied so that the model ventricle geometries under end-of-systole pressure matched in vivo data. Our results indicated that the modeling approach has the potential to be used to determine ventricle material properties. The equivalent Young's modulus value from the healthy LV (LV1) was about 30% softer than that of the infarct LV (LV2) at end of diastole, but was about 100% stiffer than that of LV2 at end of systole. This can be explained as LV1 has more active contraction reflected by stiffness variations. Using averaged values, at end-systole, longitudinal curvature from LV2 was 164% higher than that from LV1. LV stress from LV2 was 82% higher than that from LV1. At end-diastole, L-curvature from LV2 was still 132% higher than that from LV1, while LV stress from LV2 was only 9% higher than that from LV1. Longitudinal curvature and stress showed the largest differences between the two ventricles, with the LV with infarct having higher longitudinal curvature and stress values. Large scale studies are needed to further confirm our findings.

We present a high performance modularly-built open-source software - OpenIFEM. OpenIFEM is a C++ implementation of the modified immersed finite element method (mIFEM) to solve fluid-structure interaction (FSI) problems. This software is modularly built to perform multiple tasks including fluid dynamics (incompressible and slightly compressible fluid models), linear and nonlinear solid mechanics, and fully coupled fluid-structure interactions. Most of open-source software packages are restricted to certain discretization methods; some are under-tested, under-documented, and lack modularity as well as extensibility. OpenIFEM is designed and built to include a set of generic classes for users to adapt so that any fluid and solid solvers can be coupled through the FSI algorithm. In addition, the package utilizes well-developed and tested libraries. It also comes with standard test cases that serve as software and algorithm validation. The software can be built on cross-platform, i.e., Linux, Windows, and Mac OS, using CMake. Efficient parallelization is also implemented for high-performance computing for large-sized problems. OpenIFEM is documented using Doxygen and publicly available to download on GitHub. It is expected to benefit the future development of FSI algorithms and be applied to a variety of FSI applications.