Search

BACKGROUND: Registration of three-dimensional contrast-enhanced ultrasound fusion imaging (3DCEUS-FI) is time-consuming to obtain high success rate. OBJECTIVE: To investigate the influence factors on registration success rate of 3DCEUS-FI. METHODS: Water tank phantoms were made to obtain mimicked pre- and post- radiofrequency ablation three-dimensional contrast-enhanced ultrasound (3DCEUS) and CT images. Orthogonal trials were designed according to factors including size, depth, enhancement level of mimicked tumor, diameter and number of mimicked adjacent vessels. Mimicked pre- and post-RFA 3DCEUS images of 72 trials were fused to assess ablative margin (AM) by two radiologists. With CT images as standard, 3DCEUS-FI accuracy was considered as the consistency of AM evaluation. The inter-observer agreement and the influence factors on registration success rates were analyzed. RESULTS: The intraclass correlation coefficient (ICC) for the consistency of AM evaluation between CT and 3DCEUS-FI in x-axis, y-axis or z-axis was 0.840∼0.948 (P

To establish a breast lesion risk stratification system using ultrasound images to predict breast malignancy and assess Breast Imaging Reporting and Data System (BI-RADS) categories simultaneously.This multicenter study prospectively collected a dataset of ultrasound images for 5012 patients at thirty-two hospitals from December 2018 to December 2020. A deep learning (DL) model was developed to conduct binary categorization (benign and malignant) and BI-RADS categories (2, 3, 4a, 4b, 4c, and 5) simultaneously. The training set of 4212 patients and the internal test set of 416 patients were from thirty hospitals. The remaining two hospitals with 384 patients were used as an external test set. Three experienced radiologists performed a reader study on 324 patients randomly selected from the test sets. We compared the performance of the DL model with that of three radiologists and the consensus of the three radiologists.In the external test set, the DL model achieved areas under the receiver operating characteristic curve (AUCs) of 0.980 and 0.945 for the binary categorization and six-way categorizations, respectively. In the reader study set, the DL BI-RADS categories achieved a similar AUC (0.901 vs. 0.933, p = 0.0632), sensitivity (90.98% vs. 95.90%, p = 0.1094), and accuracy (83.33% vs. 79.01%, p = 0.0541), but higher specificity (78.71% vs. 68.81%, p = 0.0012) than those of the consensus of the three radiologists.The DL model performed well in distinguishing benign from malignant breast lesions and yielded outcomes similar to experienced radiologists. This indicates the potential applicability of the DL model in clinical diagnosis.• The DL model can achieve binary categorization for benign and malignant breast lesions and six-way BI-RADS categorizations for categories 2, 3, 4a, 4b, 4c, and 5, simultaneously. • The DL model showed acceptable agreement with radiologists for the classification of breast lesions. • The DL model performed well in distinguishing benign from malignant breast lesions and had promise in helping reduce unnecessary biopsies of BI-RADS 4a lesions.

A new method for the accurate discrete technique of directional spectra is proposed for the simulation of the second-order incident potential of directional seas, and a second-order time-domain analysis program is developed for nonlinear wave forces on marine structures in multidirectional irregular seas. The boundary-value problems for the diffracted potential are solved by a higher-order boundary element method (HOBEM). The fourth-order Adams–Bashforth–Moulton method combined with the Newton–Raphson iteration is implemented for time integration. The present method is validated by comparison with the published frequency-domain results for a bottom-mounted vertical cylinder in bichromatic bidirectional waves. The second-order diffracted forces on a truncated cylinder are examined for different combinations of storms and swell seas, and major efforts are focused on investigating the effects of wave directionality, directional spreading, and the spectral peak frequency of mixed seas.

BackgroundStudies on deep learning (DL)-based models in breast ultrasound (US) remain at the early stage due to a lack of large datasets for training and independent test sets for verification. We aimed to develop a DL model for differentiating benign from malignant breast lesions on US using a large multicenter dataset and explore the model’s ability to assist the radiologists.MethodsA total of 14,043 US images from 5012 women were prospectively collected from 32 hospitals. To develop the DL model, the patients from 30 hospitals were randomly divided into a training cohort (n = 4149) and an internal test cohort (n = 466). The remaining 2 hospitals (n = 397) were used as the external test cohorts (ETC). We compared the model with the prospective Breast Imaging Reporting and Data System assessment and five radiologists. We also explored the model’s ability to assist the radiologists using two different methods.ResultsThe model demonstrated excellent diagnostic performance with the ETC, with a high area under the receiver operating characteristic curve (AUC, 0.913), sensitivity (88.84%), specificity (83.77%), and accuracy (86.40%). In the comparison set, the AUC was similar to that of the expert (p = 0.5629) and one experienced radiologist (p = 0.2112) and significantly higher than that of three inexperienced radiologists (p ConclusionsThe DL model yielded satisfactory predictions in distinguishing benign from malignant breast lesions. The model showed the potential value in improving the diagnosis of breast lesions by radiologists.

Background: Graf’s method is currently the most commonly used ultrasound-based technique for the diagnosis of developmental dysplasia of the hip (DDH). However, the efficiency and accuracy of diagnosis are highly affected by the sonographers’ qualification and the time and effort expended, which has a significant intra- and inter-observer variability. Methods: Aiming to minimize the manual intervention in the diagnosis process, we developed a deep learning-based computer-aided framework for the DDH diagnosis, which can perform fully automated standard plane detection and angle measurement for Graf type I and type II hips. The proposed framework is composed of three modules: an anatomical structure detection module, a standard plane scoring module, and an angle measurement module. This framework can be applied to two common clinical scenarios. The first is the static mode, measurement and classification are performed directly based on the given standard plane. The second is the dynamic mode, where a standard plane from ultrasound video is first determined, and measurement and classification are then completed. To the best of our knowledge, our proposed framework is the first CAD method that can automatically perform the entire measurement process of Graf’s method. Results: In our experiments, 1051 US images and 289 US videos of Graf type I and type II hips were used to evaluate the performance of the proposed framework. In static mode, the mean absolute error of α, β angles are 1.71° and 2.40°, and the classification accuracy is 94.71%. In dynamic mode, the mean absolute error of α, β angles are 1.97° and 2.53°, the classification accuracy is 89.51%, and the running speed is 31 fps. Conclusions: Experimental results demonstrate that our fully automated framework can accurately perform standard plane detection and angle measurement of an infant’s hip at a fast speed, showing great potential for clinical application.

To investigate the feasibility of assessing the ablative margin (AM) of radiofrequency ablation (RFA) for hepatocellular carcinoma (HCC) with 3-D contrast-enhanced ultrasound fusion imaging (3-DCEUS-FI), pre- and post-RFA 3-DCEUS images of 84 patients with HCC were fused for two radiologists to independently assess the AMs. The success rate, duration and influencing factors for registration; inter-observer agreement for AM classification; and local tumor progression (LTP) rate were evaluated. The success rate of the automatic registration (AR), which was completed within 4-12 s, was 57.1% (48/84). The duration and success rate of the interactive registration (IR) were 4.2 ± 1.8 min and 91.7% (77/84) for radiologist A and 4.8 ± 2.1 min and 91.7% (77/84) for radiologist B, respectively. The multivariate analysis demonstrated that the pre-RFA image quality, number of vessels (≥3 mm) and presence of acoustic shadow were independent factors for AR (p0.05), while the number of vessels was an independent factor for IR (p = 0.001). The agreement between observers was excellent (kappa = 0.914). LTP rate was significantly higher for AMs5 mm than for AMs ≥5 mm (p = 0.024). Quantitatively evaluating the AM immediately after RFA for HCC with 3-DCEUS-FI was feasible.

Objective: Esophageal carcinosarcoma (ECS) is a rare malignant tumor that accounts for only 0.5%-2.8% of all esophageal malignancies. As most current studies are case reports, the relationship between clinical features and prognosis remains controversial. Methods: We investigated the clinical features and prognosis of 24 patients with ECS in a single center from 2006 to 2018. There were 18 male and 6 female patients aged 52-82 years with a median age of 62.5 years. In addition, we included 9 studies on ECS from PubMed and a literature review. Results: The median follow-up time of the 24 patients was 70.5 (range, 10-156)months. The 3-year and 5-year survival rates were 83.3% and 70.8%, respectively. Among the 24 patients, none of the 10 (41.7%) stage T1 cancer patients had lymph node metastasis; however, lymph node metastasis was noted in 8 (57.1%) stage T2-4 cancer patients. The literature review revealed that 211 patients had a 5-year survival rate of 11.8%-68.2%, and 54.5%-95.8% study participants had early stage ECS. Although the information provided in the literature review is limited, it appears to be a characteristic of the early stage of the disease and predicts better prognosis when ECS is diagnosed, which is similar to the result of the current study. Conclusion: Our results indicate that ECS has a favorable prognosis, even among patients with early stage ECS who undergo radical esophagectomy with lymph node dissection. Because of the low incidence of ECS, further studies with more cases need to investigate this rare malignancy.

PurposeUltrasound (US)-magnetic resonance (MR) fusion imaging is a profitable tool for image-guided abdominal diagnosis and biopsy. However, the automatic registration of liver US and MR images remains a challenging task. An effective local structure orientation descriptor (LSOD) for use in registering multimodal images is proposed in this study. MethodsLSOD utilizes a normalized similarity distance vector of intra-image patch pairs to extract intensity change orientations from intensity value changes in a local area. The multimodal similarity measure is then derived using the LSOD vector difference. Experiments were performed on simulated US and liver 2D US-3D MR images from a phantom, two healthy volunteers, and seven patients. ResultsUsing the LSOD-based method, the root-mean-square target registration errors (RMS-TREs) were 1.76ź1.90mm/2.03ź0.84mm in phantom/clinical experiments. All of the results outperformed those obtained using modality independent neighborhood descriptor (MIND)- and linear correlation of linear combination (LC2)-based methods (phantom/clinical: 5.23ź3.35mm/4.32ź3.63mm and 9.79ź5.03mm/6.29ź3.85mm, respectively). The registration cover range for all subjects of the LSOD-based method was 9.16mm, which was larger than those of the MIND- and LC2-based methods (5.06 and 5.12mm, respectively). ConclusionsThe results demonstrated that the LSOD-based registration method could robustly register 2D US and 3D MR images of different liver sections with acceptable accuracy for clinical requirements. This approach is useful for the practical clinical application of the US-MR fusion imaging technique. A local structure orientation descriptor for multimodal registration is proposed.Intra-image similarity is used to describe local structure in high dimensional space.Normalized similarity distance vector is derived to extract structure orientation.The method shows credible performance to register liver US and MR images.It is useful for improve the clinical practicability of US fusion imaging.

A twice expansion method in the time-domain was proposed for computing wave interaction with a body experiencing large amplitude drift motion. The body displacement was divided into two components, a low frequency and a high frequency oscillations. The low frequency oscillation was computed by a wavelet transform method with time matching in the computation, and was called as the instantaneous mean position. Then the first and the second order boundary conditions were established at the instantaneous mean position and the still free surface by a perturbation expansion method which would guarantee that the expansion factors are always small. The velocity potential was separated into the incident, the scattering and the steady disturbance potentials. The scattering and the steady disturbance potentials were computed by solving integral equations respectively. Numerical examinations were carried out for a forced moving truncated cylinder in waves and wave interaction with a spar platform with mooring system in deep water. The comparisons were made between the results from the present twice expansion method and the direct expansion method. The comparisons show that the present method can consider the changes in the phase and the encounter frequency of wave loads due to body drift motion at low frequency. The body response at the natural frequency of the spar platform from the twice expansion method is smaller than that from the direct expansion method, but is larger at lower frequency. The peak frequency of the response amplitude from the present method moves to lower frequency, as the influence of wave drift damping is included in the model.

Hydrodynamic behavior of two-dimensional tandem-arranged flapping flexible foils in uniform flow is investigated numerically by an immersed boundary-lattice Boltzmann method. The leading edge of the leading foil is forced to undergo both heave and pitch motions, while the leading edge of the trailing foil is forced to undergo heave motion only. Of particular interests are the effects of stream-wise gap distance Gx (Gx/c = 0.25–1.75, where c denotes the length of the foil) and the phase difference Φ between the heave motions of the foils (Φ/π = −1.00 to 1.00) on the hydrodynamic characteristics of the foils, such as the propulsive force, the propulsive efficiency, the passive deformation, and the flow field around the foils. For the leading foil, because of the existence of the trailing foil and the resulting gap flow between the foils, the propulsive performance is noticeably influenced by Φ at small Gx/c values and such an influence is weakened with increasing Gx/c. For the trailing foil, the propulsive performance is primarily affected by Φ, and the physics behind such a strong effect is that Φ dictates the manner by which the vortices shed from the leading foil interact with the trailing foil. In contrast, the interaction of the vortices shed from the leading foil with the trailing foil is not significantly affected by Gx/c because the trailing foil experiences similar vortices shed from the leading foil, regardless of Gx/c. With different Gx/c and Φ/π values, three distinct deformation states of the foils, namely, the symmetric periodic state, the asymmetric periodic state, and the irregular state, are identified and are mapped out in the (Φ/π, Gx/c) space. Good correlation between the deformation state of the foils and the propulsive performance of the trailing foil has been observed.

To verify the value of deep learning in diagnosing nonalcoholic fatty liver disease (NAFLD) by comparing 3 image-processing techniques.A total of 240 participants were recruited and divided into 4 groups (normal, mild, moderate, and severe NAFLD groups), according to the definition and the ultrasound scoring system for NAFLD. Two-dimensional hepatic imaging was analyzed by the envelope signal, grayscale signal, and deep-learning index obtained by 3 image-processing techniques. The values of the 3 methods ranged from 0 to 65,535, 0 to 255, and 0 to 4, respectively. We compared the values among the 4 groups, draw receiver operating characteristic curves, and compared the area under the curve (AUC) values to identify the best image-processing technique.The envelope signal value, grayscale value, and deep-learning index had a significant difference between groups and increased with the severity of NAFLD (P .05). The 3 methods showed good ability (AUC 0.7) to identify NAFLD. Meanwhile, the deep-learning index showed the superior diagnostic ability in distinguishing moderate and severe NAFLD (AUC = 0.958).The envelope signal and grayscale values were vital parameters in the diagnosis of NAFLD. Furthermore, deep learning had the best sensitivity and specificity in assessing the severity of NAFLD.

A Level-set and Finite Element Method (FEM) based two-dimensional numerical model is proposed for the free surface flows in the frame of potential theory. The FEM is used to solve the governing equation of potential flow while the Level-set method is used to capture the free surface. A new numerical technique is proposed to treat the partial elements that are cut by the free surface. Fully nonlinear boundary condition is used on the free surface. During the simulation, Eulerian meshes are adopted and only the fluid phase is calculated. A linear sloshing problem is firstly used as a benchmark to check the accuracy of the FEM solver. The numerical results are compared with the theoretical solution, and good agreements are obtained. Then the dam breaking flow with violent free surface is used to validate the proposed FEM-Level-set model. The comparisons show that present model has good numerical stability, and is able to produce reasonable prediction on the free surface evolution. Finally a fully nonlinear sloshing problem is simulated. The numerical results of the present model agree well with those obtained by the model based on boundary element method.

This paper presents an embedding framework for myocardial velocity processing with MRI based on a hollow, semi-spherical template. The relationship between vector bundle and manifold mapping is analysed and three different mapping methods that include tangent mapping, normal mapping, and normal-tangent mapping are assessed for their practical value of myocardial contractility analysis. The proposed method provides a basis for consistent volume matching and vector correspondence, in addition to the ease of calculating biomechanical indices such as radial, circumferential and longitudinal strain rates without the concern of boundary effects. Detailed analysis results with both synthetic and in vivo MR velocity data sets are provided.

In view of the non-steady of temperature field in the large oil-immersed self-cooled three-phrase transformer, this paper constructs parameter identification model and make the numerical simulation. Due to the nonsmooth property of the temperature field of the transformer, the overlapping domain decomposition method is employed to make the coil under the cylindrical coordinate, and the core limb, the yoke and the oil out of the coil under the rectangular coordinate. The interface conditions among subdomains are given. Finally the optimal algorithm is proposed to solve the parameter identification problem and the corresponding software is made. The comparison of experimental data and numerical results shows the correctness and efficiency of our methods.