Lung mass density analysis using deep neural network and lung ultrasound surface wave elastography

Weight retention and expansion of popular lead-based and lead-free hunting bullets

Hunting bullets are often comprised of a lead core covered with a copper alloy jacket. When the bullet collides with an animal, particles-sometimes millions-can shed from the projectile and embed in animal tissues. Those lead fragments can persist in game meat and remain in the discarded viscera that many wildlife species scavenge. Bullets often differ in design, so it is vital to assess their weight retention and expansion, which affects how much metal they deposit in tissue and how effectively they kill animals. We fired 12 types of hunting bullets into water to measure their weight retention and expansion at 91 m and 238 m (100 and 260 yards). Bullet constructions included copper, tin, bonded lead, partitioned lead, and cup-and-core lead. On average, copper bullets retained >98 % of their weight, whereas cup-and-core lead bullets retained <13-55 %, depending on the brand and shot distance. One brand of bonded lead bullet retained mass (≥96 %) nearly as well as copper bullets, while another brand retained much less (~71 %). Two types of copper bullets expanded similarly between test distances, while a third expanded less at 238 m. Cup-and-core lead bullets often experienced a separation between their copper alloy jacket and lead core. Our data emphasize that lead-based bullets of similar construction can drastically differ in weight retention and expansion.

Development of Biopsy Tract Sealants Based on Shape Memory Polymer Foams

Lung biopsies are often used to aid in the diagnosis of cancers. However, the procedure carries the dual risk of air (pneumothorax) or blood (hemothorax) filling the pleural cavity, increasing the risk of a collapsed lung and chest intubation. This work demonstrates the effectiveness of a polyurethane-based shape memory polymer foam as a biopsy tract sealant. The impact of diameter, length, pore size, and shape memory effect was evaluated to determine the ideal device design for tract sealing. Characterization in an in vitro benchtop lung model identified that diameter had the largest influence on sealing efficacy, while the length of the device had little to no impact. Finally, evaluation of deployment force demonstrated that devices fabricated from the shape memory polymer foams were easier to deploy than elastic foams. Following characterization, down-selected device designs were combined with radiopaque markers for use in image-guided based procedures. Furthermore, the introduction of the markers or sterilization did not impact the ability of the devices to seal the biopsy tract and led to a decrease in the deployment force. Overall, these results demonstrate the potential for polyurethane-based shape memory foam devices to serve as biopsy tract sealant devices that aim to reduce complications, such as pneumothorax, from occurring.

Artificial intelligence - based ultrasound elastography for disease evaluation - a narrative review

Ultrasound elastography (USE) provides complementary information of tissue stiffness and elasticity to conventional ultrasound imaging. It is noninvasive and free of radiation, and has become a valuable tool to improve diagnostic performance with conventional ultrasound imaging. However, the diagnostic accuracy will be reduced due to high operator-dependence and intra- and inter-observer variability in visual observations of radiologists. Artificial intelligence (AI) has great potential to perform automatic medical image analysis tasks to provide a more objective, accurate and intelligent diagnosis. More recently, the enhanced diagnostic performance of AI applied to USE have been demonstrated for various disease evaluations. This review provides an overview of the basic concepts of USE and AI techniques for clinical radiologists and then introduces the applications of AI in USE imaging that focus on the following anatomical sites: liver, breast, thyroid and other organs for lesion detection and segmentation, machine learning (ML) - assisted classification and prognosis prediction. In addition, the existing challenges and future trends of AI in USE are also discussed.

State of the Art in Lung Ultrasound, Shifting from Qualitative to Quantitative Analyses

Lung ultrasound (LUS) has been increasingly expanding since the 1990s, when the clinical relevance of vertical artifacts was first reported. However, the massive spread of LUS is only recent and is associated with the coronavirus disease 2019 (COVID-19) pandemic, during which semi-quantitative computer-aided techniques were proposed to automatically classify LUS data. In this review, we discuss the state of the art in LUS, from semi-quantitative image analysis approaches to quantitative techniques involving the analysis of radiofrequency data. We also discuss recent in vitro and in silico studies, as well as research on LUS safety. Finally, conclusions are drawn highlighting the potential future of LUS.

Semantic Analysis Technology of English Translation Based on Deep Neural Network

English translation plays an important role in the development of science and technology and cultural exchanges. With the increase in translation volume, intelligent translation has become inevitable, but there is no effective solution for semantic translation in English translation. To provide an effective translation improvement scheme in English translation, this paper studies and analyzes the application of deep neural network in English translation semantic analysis. Based on a brief analysis of the research progress of English translation analysis and the current situation of neural network, a neural network translation architecture is established, and a deep neural network model for English translation analysis is proposed. Aiming at the problem of gradient disappearance in RNN model, the ability of Gru neural network to deal with long-distance translation is enhanced, and the computational complexity of Gru neural network is reduced. At the same time, a bi-directional Gru model is designed to translate according to the context. For some nonlinear translation, a deep neural network model based on part of speech sequence information is proposed to realize semantic analysis, and experiments are designed to test the translation effect of the neural network model. The simulation results show that the English translation based on deep neural network can improve the translation effect, reduce errors and improve the accuracy of semantic analysis of English translation, which has certain reference significance for improving the level of English translation.

Artificial Intelligence in Quantitative Ultrasound Imaging: A Survey

Quantitative ultrasound (QUS) imaging is a safe, reliable, inexpensive, and real-time technique to extract physically descriptive parameters for assessing pathologies. Compared with other major imaging modalities such as computed tomography and magnetic resonance imaging, QUS suffers from several major drawbacks: poor image quality and inter- and intra-observer variability. Therefore, there is a great need to develop automated methods to improve the image quality of QUS. In recent years, there has been increasing interest in artificial intelligence (AI) applications in medical imaging, and a large number of research studies in AI in QUS have been conducted. The purpose of this review is to describe and categorize recent research into AI applications in QUS. We first introduce the AI workflow and then discuss the various AI applications in QUS. Finally, challenges and future potential AI applications in QUS are discussed.

Ultrasound deep learning for monitoring of flow-vessel dynamics in murine carotid artery

Several arterial diseases are closely related with mechanical properties of the blood vessel and interactions of flow-vessel dynamics such as mean flow velocity, wall shear stress (WSS) and vascular strain. However, there is an opportunity to improve the measurement accuracy of vascular properties and hemodynamics by adopting deep learning-based ultrasound imaging for flow-vessel dynamics (DL-UFV). In this study, the DL-UFV is proposed by devising an integrated neural network for super-resolved localization and vessel wall segmentation, and it is also combined with tissue motion estimation and flow measurement techniques such as speckle image velocimetry and speckle tracking velocimetry for measuring velocity field information of blood flow. Performance of the DL-UFV is verified by comparing with other conventional techniques in tissue-mimicking phantoms. After the performance verification, in vivo feasibility is demonstrated in the murine carotid artery with different pathologies: aging and diabetes mellitus (DM). The mutual comparison of flow-vessel dynamics and histological analyses shows correlations between the immunoreactive region and abnormal flow-vessel dynamics interactions. The DL-UFV improves biases in measurements of velocity, WSS, and strain with up to 4.6-fold, 15.1-fold, and 22.2-fold in the tissue-mimicking phantom, respectively. Mean flow velocities and WSS values of the DM group decrease by 30% and 20% of those of the control group, respectively. Mean flow velocities and WSS values of the aging group (34.11 cm/s and 13.17 dyne/cm²) are slightly smaller than those of the control group (36.22 cm/s and 14.25 dyne/cm²). However, the strain values of the aging and DM groups are much smaller than those of the control group (p < 0.05). This study shows that the DL-UFV performs better than the conventional ultrasound-based flow and strain measurement techniques for measuring vascular stiffness and complicated flow-vessel dynamics. Furthermore, the DL-UFV demonstrates its excellent performance in the analysis of the hemodynamic and hemorheological effects of DM and aging on the flow and vascular characteristics. This work provides useful hemodynamic information, including mean flow velocity, WSS and strain with high-resolution for diagnosing the pathogenesis of arterial diseases. This information can be used for monitoring progression and regression of atherosclerotic diseases in clinical practice.

Lung Ultrasound Surface Wave Elastography for Assessing Patients With Pulmonary Edema

B-Mode ultrasound insonation of lungs that are dense with extravascular lung water (EVLW) produces characteristic reverberation artifacts termed B-lines. The number of B-lines present demonstrates reasonable correlation to the amount of EVLW. However, analysis of B-line artifacts generated by this modality is semi-quantitative relying on visual interpretation, and as a result, can be subject to inter-observer variability. The purpose of this study was to translate the use of a novel, quantitative lung ultrasound surface wave elastography technique (LUSWE) into the bedside assessment of pulmonary edema in patients admitted with acute congestive heart failure. B-mode lung ultrasound and LUSWE assessment of the lungs were performed using anterior and lateral intercostal spaces in the supine patient. 14 patients were evaluated at admission with reassessment performed 1-2 days after initiation of diuretic therapy. Each exam recorded the total lung B-lines, lung surface wave speeds (at 100, 150, and 200 Hz) and net fluid balance. The patient cohort experienced effective diuresis (average net fluid balance of negative 2.1 liters) with corresponding decrease in pulmonary edema visualized by B-mode ultrasound (average decrease of 13 B-Lines). In addition, LUSWE demonstrated a statistically significant reduction in the magnitude of wave speed from admission to follow-up. The reduction in lung surface wave speed suggests a decrease in lung stiffness (decreased elasticity) mediated by successful reduction of pulmonary edema. In summary, LUSWE is a noninvasive technique for quantifying elastic properties of superficial lung tissue that may prove useful as a diagnostic test, performed at the bedside, for the quantitative assessment of pulmonary edema.

Lung mass density prediction using machine learning based on ultrasound surface wave elastography and pulmonary function testing

OBJECTIVE

The objective of this study is to predict in vivo lung mass density for patients with interstitial lung disease using different gradient boosting decision tree (GBDT) algorithms based on measurements from lung ultrasound surface wave elastography (LUSWE) and pulmonary function testing (PFT).

METHODS

Age and weight of study subjects (57 patients with interstitial lung disease and 20 healthy subjects), surface wave speeds at three vibration frequencies (100, 150, and 200 Hz) from LUSWE, and predicted forced expiratory volume (FEV1% pre) and ratio of forced expiratory volume to forced vital capacity (FEV1%/FVC%) from PFT were used as inputs while lung mass densities based on the Hounsfield Unit from high resolution computed tomography (HRCT) were used as labels to train the regressor in three GBDT algorithms, XGBoost, CatBoost, and LightGBM. 80% (20%) of the dataset was used for training (testing).

RESULTS

The results showed that predictions using XGBoost regressor obtained an accuracy of 0.98 in the test dataset.

CONCLUSION

The obtained results suggest that XGBoost regressor based on the measurements from LUSWE and PFT may be able to noninvasively assess lung mass density in vivo for patients with pulmonary disease.

Ultrasound Elastography for Lung Disease Assessment

Ultrasound elastography (US-E) is a noninvasive, safe, cost-effective and reliable technique to assess the mechanical properties of soft tissue and provide imaging biomarkers for pathological processes. Many lung diseases such as acute respiratory distress syndrome, chronic obstructive pulmonary disease, and interstitial lung disease are associated with dramatic changes in mechanical properties of lung tissues. Nevertheless, US-E is rarely used to image the lung because it is filled with air. The large difference in acoustic impedance between air and lung tissue results in the reflection of the ultrasound wave at the lung surface and, consequently, the loss of most ultrasound energy. In recent years, there has been an increasing interest in US-E applications in evaluating lung diseases. This article provides a comprehensive review of the technological advances of US-E research on lung disease diagnosis. We introduce the basic principles and major techniques of US-E and provide information on various applications in lung disease assessment. Finally, the potential applications of US-E to the diagnosis of COVID-19 pneumonia is discussed.

Deep Convolutional Neural Networks for Thyroid Tumor Grading using Ultrasound B-mode Images

The performances of deep convolutional neural network (DCNN) modeling and transfer learning (TF) for thyroid tumor grading using ultrasound imaging were evaluated. This retrospective study included input patient data (ultrasound B-mode image sets) assigned to the training group (115 participants) or testing group (28 participants). DCNN (ResNet50) and TF (ResNet50, ResNet101, ResNet152, VGG16, Inception V3, and DenseNet201), which trains a convolutional neural network that has been pre-trained on ImageNet, were used for image classification based on thyroid tumor grade. Supervised training was performed by using the DCNN or TF model to minimize the difference between the output data and clinical grading. The performances of the DCNN and TF models were assessed in the testing dataset with receiver operating characteristic analyses. Results showed that TF based on Resnet50 and VGG16 had better performance than DCNN (ResNet50) in differentiating thyroid tumor with areas under the receiver operating characteristic (AUCs) curve more than 0.8. However, TF based on ResNet101, ResNet152, InceptionV3, and Densenet201 had equal or worse performances than DCNN (ResNet50) in grading thyroid tumor with AUCs less than 0.5. TF based on ResNet50 and VGG16 had a superior performance compared to DCNN (ResNet50) model for grading thyroid tumors based on ultrasound images.

Predicting lung mass density of patients with interstitial lung disease and healthy subjects using deep neural network and lung ultrasound surface wave elastography

The Hounsfield unit (HU) obtained from high resolution computed tomography (HRCT) has been used to assess lung pathology. However, lung mass density has not been quantified in vivo noninvasively. The objective of this study was to develop a method for analyzing lung mass density of superficial lung tissue of patients with interstitial lung disease (ILD) and healthy subjects using a deep neural network (DNN) and lung ultrasound surface wave elastography (LUSWE). Surface wave speeds at three vibration frequencies (100, 150 and 200 Hz) from LUSWE and a pulmonary function test (PFT) including predicted forced expiratory volume (FEV1% pre) and ratio of forced expiratory volume to forced vital capacity (FEV1%/FVC%) were used. Predefined lung mass densities based on the HU for ILD patients and healthy subjects (77 in total) were also used to train the DNN model. The DNN was composed of four hidden layers of 1024 neurons for each layer and trained for 80 epochs with a batch size of 20. The learning rate was 0.001. Performances of two types of activation functions in the DNN, rectified linear activation unit (ReLU) and exponential linear unit (ELU), as well as, machine learning models (support vector regression, random forest, Adaboost) were evaluated. The test dataset of wave speeds, FEV1% pre and FEV%/FVC%, was used to predict lung mass density. The results showed that predictions using a DNN with ELU obtained a comparatively better performance in the testing dataset (accuracy = 0.89) than those of DNN with ReLU or machine learning models. This method may be useful to noninvasively analyze lung mass density by using the DNN model together with the measurements from LUSWE and PFT.

OCTRexpert:A Feature-based 3D Registration Method for Retinal OCT Images

Medical image registration can be used for studying longitudinal and cross-sectional data, quantitatively monitoring disease progression and guiding computer assisted diagnosis and treatments. However, deformable registration which enables more precise and quantitative comparison has not been well developed for retinal optical coherence tomography (OCT) images. This paper proposes a new 3D registration approach for retinal OCT data called OCTRexpert. To the best of our knowledge, the proposed algorithm is the first full 3D registration approach for retinal OCT images which can be applied to longitudinal OCT images for both normal and serious pathological subjects. In this approach, a pre-processing method is first performed to remove eye motion artifact and then a novel design-detection-deformation strategy is applied for the registration. In the design step, a couple of features are designed for each voxel in the image. In the detection step, active voxels are selected and the point-to-point correspondences between the subject and template images are established. In the deformation step, the image is hierarchically deformed according to the detected correspondences in multi-resolution. The proposed method is evaluated on a dataset with longitudinal OCT images from 20 healthy subjects and 4 subjects diagnosed with serious Choroidal Neovascularization (CNV). Experimental results show that the proposed registration algorithm consistently yields statistically significant improvements in both Dice similarity coefficient and the average unsigned surface error compared with the other registration methods.

Artificial neural network to estimate micro-architectural properties of cortical bone using ultrasonic attenuation: A 2-D numerical study

The goal of this study is to estimate micro-architectural parameters of cortical porosity such as pore diameter (φ), pore density (ρ) and porosity (ν) of cortical bone from ultrasound frequency dependent attenuation using an artificial neural network (ANN). First, heterogeneous structures with controlled pore diameters and pore densities (mono-disperse) were generated, to mimic simplified structure of cortical bone. Then, more realistic structures were obtained from high resolution CT scans of human cortical bone. 2-D finite-difference time-domain simulations were conducted to calculate the frequency-dependent attenuation in the 1-8 MHz range. An ANN was then trained with the ultrasonic attenuation at different frequencies as the input feature vectors while the output was set as the micro-architectural parameters (pore diameter, pore density and porosity). The ANN is composed of three fully connected dense layers with 24, 12 and 6 neurons, connected to the output layer. The dataset was trained over 6000 epochs with a batch size of 16. The trained ANN exhibits the ability to predict the micro-architectural parameters with high accuracy and low losses. ANN approaches could potentially be used as a tool to help inform physics-based modelling of ultrasound propagation in complex media such as cortical bone. This will lead to the solution of inverse-problems to retrieve bone micro-architectural parameters from ultrasound measurements for the non-invasive diagnosis and monitoring osteoporosis.

A Pilot Study of Wet Lung Using Lung Ultrasound Surface Wave Elastography in an Ex Vivo Swine Lung Model

Extravascular lung water (EVLW) is a basic symptom of congestive heart failure and other conditions. Computed tomography (CT) is standard to assess EVLW, but it requires ionizing radiation and radiology facilities. Lung ultrasound reverberation artifacts called B-lines have been used to assess EVLW. However, B-line artifact analysis relies on visual interpretation and subjects to inter-observer variability. We developed lung ultrasound surface wave elastography (LUSWE) to measure lung surface wave speed. This research aims to develop LUSWE to measure the change of lung surface wave speed due to lung water in an ex vivo swine lung model. The surface wave speeds of a fresh ex vivo swine lung were measured at four frequencies of 100 Hz, 200 Hz, 300 Hz, and 400 Hz. An amount of water was then filled into the lung through its trachea. Ultrasound imaging was used to guide the water filling until significant changes were visible on the imaging. The lung surface wave speeds were measured. An additional 120 ml of water was then filled into the lung. The lung surface wave speeds were then measured again. The results demonstrated that the lung surface wave speed decreased with respect to water content.

Predicting Structural Properties of Cortical Bone by Combining Ultrasonic Attenuation and an Artificial Neural Network (ANN): 2-D FDTD Study

The goal of this paper is to predict the micro-architectural parameters of cortical bone such as pore diameter ( ϕ ) and porosity ( v ) from ultrasound attenuation measurements using an artificial neural network (ANN). Slices from a 3-D CT scan of human femur are obtained. The micro-architectural parameters of porosity such as average pore size and porosity are calculated using image processing. When ultrasound waves propagate in porous structures, attenuation is observed due to scattering. Two-dimensional finite-difference time-domain simulations are carried out to obtain frequency dependent attenuation in those 2D structures. An artificial neural network (ANN) is then trained with the input feature vector as the frequency dependent attenuation and output as pore diameter ( ϕ ) and porosity ( v ) . The ANN is composed of one input layer, 3 hidden layers and one output layer, all of which are fully connected. 340 attenuation data sets were acquired and trained over 2000 epochs with a batch size of 32. Data was split into train, validation and test. It was observed that the ANN predicted the micro-architectural parameters of the cortical bone with high accuracies and low losses with a minimum R2 (goodness of fit) value of 0.95. ANN approaches could potentially help inform the solution of inverse-problems to retrieve bone porosity from ultrasound measurements. Ultimately, those inverse-problems could be used for the non-invasive diagnosis and monitoring of osteoporosis.

A quantitative method for measuring the changes of lung surface wave speed for assessing disease progression of interstitial lung disease

Lung ultrasound surface wave elastography (LUSWE) is a novel non-invasive technique for measuring superficial lung tissue stiffness. The purpose of the study described here was to develop LUSWE for assessment of progression in patients with interstitial lung disease (ILD). In this study, LUSWE was used to measure changes in lung surface wave speeds at 100, 150 and 200 Hz through six intercostal lung spaces for 52 patients with ILD. The mean age was 63.1 ± 12.0 y (range: 20-85, 23 male and 29 female). The follow-up interval was 9.2 ± 3.5 mo depending on each patient's return appointment and availability. For each patient, disease progression between the baseline and follow-up tests was evaluated clinically using a 7-point Likert scale comprising three grades of improvement (mild, moderate, marked), unchanged status and three grades of worsening (mild, moderate, marked). Clinical assessments were based on changes in pulmonary function tests together with high-resolution computed tomography, echocardiography and clinical evaluations. This study illustrates the correlations between changes in lung surface wave speed and clinical assessments. Correlations of changes in lung surface wave speed at lower lateral and posterior portions of the lung portions with clinical assessments were good. LUSWE provides quantitative global and regional changes in lung surface wave speed that may be useful for quantitative assessment of progression of ILD.

Rapid measurement of the fourth-order texture coefficient by laser ultrasonic surface acoustic waves based on a neural network expert system

In anisotropic materials, texture components and orientation density directly affect the surface acoustic waves' (SAW) velocity dispersion and SAW velocity variation. In this paper, a texture feature recognition and analysis system for a neural network is constructed based on the corresponding characteristics of texture components and orientation density and SAW velocity dispersion and variation by combining laser ultrasound technology with the partial texture analysis method, which is used for the identification and analysis of texture type and feature. At the same time, based on the relationship between surface wave velocity and the fourth-order texture coefficient, an expert system for accurate prediction of the fourth-order texture coefficient is constructed. Then, the fourth-order texture coefficients predicted by the neural network expert system (NNES) are compared with the texture coefficients measured by electron backscattered diffraction. The results show that the NNES can not only quickly identify and analyze texture features, but also accurately predict the fourth-order texture coefficients.

Lung Ultrasound Surface Wave Elastography for Assessing Interstitial Lung Disease

OBJECTIVE

Our goal is to translate lung ultrasound surface wave elastography (LUSWE) for assessing patients with interstitial lung disease (ILD) and various connective tissue diseases including systemic sclerosis (SSc).

METHODS

LUSWE was used to measure the surface wave speed of lung at 100, 150, and 200 Hz through six intercostal lung spaces for 91 patients with ILD and 30 healthy control subjects. In addition, skin viscoelasticity was measured on both forearms and upper arms for patients and controls.

RESULTS

The surface wave speeds of patients' lungs were significantly higher than those of control subjects. Patient skin elasticity and viscosity were significantly higher than those of control subjects. In dividing ILD patients into two groups, ILD with SSc patients and ILD without SSc patients, significant differences between each patient group with the control group were found for both the lung and skin. No significant differences were found between the two patient groups, although there were some differences at a few locations and at 100 Hz for skin viscoelasticity.

CONCLUSION

Significant differences of surface wave speed were found between ILD patients and healthy control subjects for both the lung and skin.

SIGNIFICANCE

LUSWE may be useful for assessing ILD and SSc and screening early stage patients.

The effect of pleural fluid layers on lung surface wave speed measurement: Experimental and numerical studies on a sponge lung phantom

Pleural effusion manifests as compression of pleural fluid on the lung parenchyma contributing to hypoxemia. Medical procedures such as drainage of plural fluid releases this compression and increases oxygenation. However, the effect of pleural effusion on the elasticity of lung parenchyma is unknown. By using lung ultrasound surface wave elastography (LUSWE) and finite element method (FEM), the effect of pleural effusion on the elasticity of superficial lung parenchyma in terms of surface wave speed measurement was evaluated in a sponge phantom study. Different thicknesses of ultrasound transmission gel used to simulated pleural fluid were inserted into a condom, which was placed between the sponge and standoff pad. A mechanical shaker was used to generate vibration on the sponge phantom at different frequencies ranging from 100 to 300 Hz while the ultrasound transducer was used to capture the motion for measurement of surface wave speed of the sponge. FEM was conducted based on the experimental setup and numerically assessed the influence of pleural effusion on the surface wave speed of the sponge. We found from FEM experiments that the influence of thickness of ultrasound transmission gel was statistically insignificant on the surface wave speed of the sponge at 100 and 150 Hz.