Parathyroid ultrasound is widely used in clinical practice and plays a crucial role in the diagnosis and treatment of parathyroid diseases. Nevertheless, ultrasound physicians frequently encounter a number of challenges and doubts in their professional practice. For this reason, Superficial Organs and Peripheral Vessels Committee of Chinese Association of Ultrasound in Medicine and Engineering has formulated the expert consensus on certain common clinical problems of parathyroid ultrasound based on the current research progress and clinical experience, in order to guide the clinical practice. This consensus describes in detail the diagnostic and interventional common problems of parathyroid ultrasound and provides in-depth discussion on related contents.

Medical image semantic segmentation is essential in computer-aided diagnosis systems. It can separate tissues and lesions in the image and provide valuable information to radiologists and doctors. The breast ultrasound (BUS) images have advantages: no radiation, low cost, portable, etc. However, there are two unfavorable characteristics: (1) the dataset size is often small due to the difficulty in obtaining the ground truths, and (2) BUS images are usually in poor quality. Trustworthy BUS image segmentation is urgent in breast cancer computer-aided diagnosis systems, especially for fully understanding the BUS images and segmenting the breast anatomy, which supports breast cancer risk assessment. The main challenge for this task is uncertainty in both pixels and channels of the BUS images. In this paper, we propose a Spatial and Channel-wise Fuzzy Uncertainty Reduction Network (SCFURNet) for BUS image semantic segmentation. The proposed architecture can reduce the uncertainty in the original segmentation frameworks. We apply the proposed method to four datasets: (1) a five-category BUS image dataset with 325 images, and (2) three BUS image datasets with only tumor category (1830 images in total). The proposed approach compares state-of-the-art methods such as U-Net with VGG-16, ResNet-50/ResNet-101, Deeplab, FCN-8s, PSPNet, U-Net with information extension, attention U-Net, and U-Net with the self-attention mechanism. It achieves 2.03%, 1.84%, and 2.88% improvements in the Jaccard index on three public BUS datasets, and 6.72% improvement in the tumor category and 4.32% improvement in the overall performance on the five-category dataset compared with that of the original U-shape network with ResNet-101 since it can handle the uncertainty effectively and efficiently.

Breast cancer is one of the most serious disease affecting women’s health. Due to low cost, portable, no radiation, and high efficiency, breast ultrasound (BUS) imaging is the most popular approach for diagnosing early breast cancer. However, ultrasound images are low resolution and poor quality. Thus, developing accurate detection system is a challenging task. In this paper, we propose a fully automatic segmentation algorithm consisting of two parts: fuzzy fully convolutional network and accurately fine-tuning post-processing based on breast anatomy constraints. In the first part, the image is pre-processed by contrast enhancement, and wavelet features are employed for image augmentation. A fuzzy membership function transforms the augmented BUS images into the fuzzy domain. The features from convolutional layers are processed using fuzzy logic as well. The conditional random fields (CRFs) post-process the segmentation result. The location relation among the breast anatomy layers is utilized to improve the performance. The proposed method is applied to the dataset with 325 BUS images, and achieves state-of-the-art performance compared with that of existing methods with true positive rate 90.33%, false positive rate 9.00%, and intersection over union (IoU) 81.29% on tumor category, and overall intersection over union (mIoU) 80.47% over five categories: fat layer, mammary layer, muscle layer, background, and tumor.

Automatic breast ultrasound image (BUS) segmentation is still a challenging task due to poor image quality and inherent speckle noise. In this paper, we propose a novel multi-scale fuzzy generative adversarial network (MSF-GAN) for breast ultrasound image segmentation. The proposed MSF-GAN consists of two networks: a generative network to generate segmentation maps for input BUS images, and a discriminative network that employs a multi-scale fuzzy (MSF) entropy module for discrimination. The major contribution of this paper is applying fuzzy logic and fuzzy entropy in the discriminative network which can distinguish the uncertainty of segmentation maps and groundtruth maps and forces the generative network to achieve better segmentation performance. We evaluate the performance of MSF-GAN on three BUS datasets and compare it with six state-of-the-art deep neural network-based methods in terms of five metrics. MSF-GAN achieves the highest mean IoU of 78.75%, 73.30%, and 71.12% on three datasets, respectively.

Convolutional neural networks (CNNs) are widely used in medical image analysis, especially for breast ultrasound (BUS) image segmentation. Automatically encoding deep features is one of the most important reasons leading to the success of deep convolutional neural networks. There are a lot of studies on obtaining better convolutional features; how-ever, they do not discuss the higher-order information in the features. In this research, we propose a novel convolutional operator, a shape-adaptive convolutional operator, which can select pixels for calculating convolution rather than in the Euclidean space. The proposed operator is combined with the original convolutional operator to extract higher-order convolutional features. We conduct extensive experiments to evaluate the performance of the proposed operator for image segmentation using three datasets: two public BUS image datasets and one multi-category BUS image dataset. The proposed approach achieves state-of-the-art performance.

Deep learning approaches have achieved impressive results in breast ultrasound (BUS) image segmentation. However, these methods did not solve uncertainty and noise in BUS images well. Meanwhile, they did not involve the context information of BUS images, either. To address this issue, we present a novel deep learning structure for BUS image semantic segmentation by analyzing the uncertainty using a pyramid fuzzy block and generating a novel feature based on connectedness. There are three major contributions in this paper: (1) the structure of pyramid fuzzy block; (2) a novel membership function based on multi-convolution layers; and (3) a novel context feature based on connectedness. The proposed methods are applied to two datasets: a BUS image benchmark with two categories (background and tumor) and a five-category BUS image dataset with fat layer, mammary layer, muscle layer, background, and tumor. The proposed method achieves the best results on both datasets compared with eight state-of-the-art deep learning-based approaches.

Hypoxia, an important component of the tumor microenvironment, plays a crucial role in the occurrence and progression of cancer. However, to the best of our knowledge, a systematic analysis of a hypoxia-related prognostic signature for breast cancer is lacking and is urgently required. Therefore, in the present study, RNA-seq data and clinical information were downloaded from The Cancer Genome Atlas (TCGA) and served as a discovery cohort. Cox proportional hazards regression analysis was performed to construct a 14-gene prognostic signature (PFKL, P4HA2, GRHPR, SDC3, PPP1R15A, SIAH2, NDRG1, BTG1, TPD52, MAFF, ISG20, LALBA, ERRFI1 and VHL). The hypoxia-related signature successfully predicted survival outcomes of the discovery cohort (P<0.001 for the TCGA dataset). Three independent Gene Expression Omnibus databases (GSE10886, GSE20685 and GSE96058) were used as validation cohorts to verify the value of the predictive signature (P=0.007 for GSE10886, P=0.021 for GSE20685, P<0.001 for GSE96058). In the present study, a robust predictive signature was developed for patients with breast cancer, and the findings revealed that the 14-gene hypoxia-related signature could serve as a potential prognostic biomarker for breast cancer.

提出一种基于深度学习的数值模式降水产品降尺度方法.利用深度学习的非线性映射能力和对栅格数据的信息提取能力,建立深度超分辨率模型提取不同分辨率数值模式降水产品间相对应的有效信息,从而将低分辨率数值模式降水产品利用提取的信息重构为高分辨率产品,继而通过构建多时次组合降尺度深度模型提取时间关联性进一步提升了重构准确性.基于欧洲中期天气预报中心不同尺度数值模式降水产品的实验表明所提方法能够比常用的双三次插值方法更有效地将低分辨率降水产品转换为对应的高分辨率产品.

Breast cancer is one of the most serious disease affects women's health. Due to low cost, portable, no radiation, and high efficiency, breast ultrasound (BUS) imaging is the most popular approach for diagnosing early breast cancer. However, ultrasound images are low resolution and poor quality, developing accurate detection system is a challenging task. In this paper, we propose a fully automatic segmentation algorithm consisting of two parts: fuzzy fully convolutional network and accurately fine-tuning post-processing based on breast anatomy constraints. In the first part, the image is preprocessed by contrast enhancement, and wavelet features are employed for image augmentation. A fuzzy membership function transforms the augmented BUS images into fuzzy domain. The features from convolutional layers are processed using fuzzy logic as well. The conditional random fields (CRFs) post-process the segmentation result. The location relation among the breast anatomy layers is utilized to improve the performance. The proposed method is applied to the dataset with 325 BUS images, and achieves state-of-art performance compared with that of existing methods with true positive rate 90.33%, false positive rate 9.00%, and intersection over union (IoU) 81.29% on tumor category, and overall intersection over union (mIoU) 80.47% over five categories: fat layer, mammary layer, muscle layer, background, and tumor.

Objective: To investigate the diagnostic performance of qualitative shear wave elastography (SWE) in the characterization of thyroid nodules, compared with quantitative SWE and grayscale ultrasound (US). Methods: A total of 174 thyroid nodules from 174 patients were preoperatively evaluated by grayscale US and SWE. The quantitative SWE metrics were measured. For the qualitative SWE analysis, a four-pattern classification was established according to the color distribution features of the elastic maps and assigned for each enrolled nodule, as 1 for no meaningful findings, 2 for a capsule-related color, 3 for marginal color, and 4 for interior color. Results: Intraobserver and inter-observer agreement of the qualitative classification proposed in this study were substantial (kappa=0.793 and 0.756, respectively). The area under the receiver operating characteristic curve (AUC) of the proposed classification was similar to that of grayscale US (P=0.73) and quantitative SWE metrics (P=0.90). Compared with the quantitative SWE, the qualitative classification yielded a higher sensitivity (P < 0.05) and a similar specificity (P=0.75). Adding SWE features to grayscale US, either qualitatively or quantitatively, improved the overall specificity (both P < 0.001). Among all data sets, the optimal diagnostic performance came from the combined set of grayscale US and the qualitative classification (AUC, 0.890), with 84.4% sensitivity and 94.6% specificity. Conclusion: The qualitative four-pattern classification proposed in this study was feasible and highly reproducible in the SWE evaluation of thyroid nodules. SWE imaging features, especially the qualitative classification, have the potential to facilitate ultrasound diagnosis for thyroid nodules.

Computer-aided diagnosis (CAD) can help doctors in diagnosing breast cancer. Breast ultrasound (BUS) imaging is harmless, effective, portable, and is the most popular modality for breast cancer detection/diagnosis. Many researchers work on improving the performance of CAD systems. However, there are two main shortcomings: (1) Most of the existing methods are based on prerequisites that there is one and only one tumor in the image. (2) The results depend on the datasets, i.e., an algorithm using different datasets may obtain different performances. It implies that the performance of traditional methods is dataset-dependent. In this paper, we propose an effective approach: (1) using information extended images to train a fully convolutional network (FCN) to semantically segment BUS image into 3 categories: mammary layer, tumor, and background; and (2) applying layer structure information - the breast cancers are located inside the mammary layer - to the conditional random field (CRF) for conducting breast cancer segmentation and making the segmentation result more accurate. The proposed method is evaluated utilizing BUS images of 325 cases, and the result is the best comparing with that of the existing methods by achieving true positive rate 92.80%, false positive rate 9%, and Intersection over Union 82.11%. The proposed approach has solved the above mentioned two shortcomings of the existing methods.

This study aimed to determine whether acoustic radiation force impulse (ARFI) can help to regulate thyroid imaging reporting and data system (TI-RADS) category 4 lesions and reassess the malignancy risk for a better selection of nodules submitted to fine-needle aspiration (FNA). ARFI was performed on 265 patients with 271 TI-RADS category 4 thyroid nodules. We calculated virtual touch tissue imaging (VTI) image grade, area ratio (VTIAR) and shear wave velocity (SWV) of lesions with VTI and virtual touch tissue quantification (VTQ). The categories of nodules were upgraded with higher-stiffness and downgraded with lower-stiffness, and the risks of malignancy changed accordingly. Surgical histology findings confirmed that 109 nodules were malignant and 162 nodules were benign. The cut-off values were VTI grade >= IV, VTIAR >= 1.12 and SWV >= 2.81 m/s, respectively. The diagnostic performance of the modified TI-RADS was statistically higher than that of TI-RADS (AUC: 0.939 vs 0.857; P < 0.05). By limiting the sizes and malignancy risks of nodules using the modified TI-RADS, the number of benign nodules which should undergo FNA could be reduced by 62 cases. ARFI is helpful to adjust the TI-RADS classification and aid doctors in making appropriate decisions in terms of FNA recommendation.

As breast cancer tissues are stiffer than normal tissues, shear wave elastography (SWE) can locally quantify tissue stiffness and provide histological information. Moreover, tissue stiffness can be observed on three-dimensional (3D) colour-coded elasticity maps. Our objective was to evaluate the diagnostic performances of quantitative features in differentiating breast masses by two-dimensional (2D) and 3D SWE. Two hundred ten consecutive women with 210 breast masses were examined with B-mode ultrasound (US) and SWE. Quantitative features of 3D and 2D SWE were assessed, including elastic modulus standard deviation (ESDE) measured on SWE mode images and ESDU measured on B-mode images, as well as maximum elasticity (Emax). Adding quantitative features to B-mode US improved the diagnostic performance (p < 0.05) and reduced false-positive biopsies (p < 0.0001). The area under the receiver operating characteristic curve (AUC) of 3D SWE was similar to that of 2D SWE for ESDE (p = 0.026) and ESDU (p = 0.159) but inferior to that of 2D SWE for Emax (p = 0.002). Compared with ESDU, ESDE showed a higher AUC on 2D (p = 0.0038) and 3D SWE (p = 0.0057). Our study indicates that quantitative features of 3D and 2D SWE can significantly improve the diagnostic performance of B-mode US, especially 3D SWE ESDE, which shows considerable clinical value.

Acoustic radiation force impulse (ARFI)-imaging is a novel ultrasound-based elastography method enabling quantitative measurement of tissue stiffness. This study aimed to evaluate diagnostic value of conventional ultrasound and tissue quantification by using acoustic radiation force impulse (ARFI) technology for differentiation of small thyroid solid lesions. Ninety thyroid masses were examined by using the conventional ultrasound and Virtual touch tissue quantification (VTQ) of ARFI. The shear wave velocity (SWV) (m/s) was also examined. Combined traditional ultrasound diagnosis criteria (CTUDC) of thyroid nodules were also evaluated. Receiver-operating characteristic curve (ROC) analysis was performed to assess the diagnostic performance. The final diagnosis was obtained from clinical histology findings. In conventional ultrasound patterns, A/T >= 1 had the highest area under curve, which could achieve to 0.6547. The mean value of SWV of thyroid microcarcinoma differed significantly from those of benign nodules (3.92 +/- 2.01 m/s vs. 2.52 +/- 1.09 m/s, P < 0.01). For differentiating between benign and malignant nodules, the sensitivity, specificity were 80.56% and 74.07%, respectively, which were based on the standard SWV (2.57 m/s). Meanwhile, the sensitivity and specificity could achieve 77.78% and 77.22%, respectively, which were based on CTUDC. The Area under ROC curve (AUC) of VTQ and CTUDC were 0.825 and 0.8226 respectively. Among the ninety thyroid masses, only five benign lesions were misdiagnosis as malignant ones if combination application of VTQ and CTUDC. In conclusion, the VTQ of ARFI technology can be applied in diagnosis of small thyroid nodules, which may be complement B-mode ultrasound and plays an important role in clinical applications.

The aim of this study was to evaluate acoustic radiation force impulse imaging for cervical lymphadenopathy in routine clinical practice and to correlate the acoustic radiation force impulse values with the morphological signs and the pathological results, which were used as the reference standard. The virtual touch tissue quantification values were analyzed in 123 patients (mean age 40.8 years, range 1–81 years) with 181 cervical lymph nodes (87 benign, 94 malignant). The diagnostic performance of acoustic radiation force impulse values were evaluated with respect to sensitivity, specificity, and area under the curve using a receiver operating characteristic curve analysis. The mean virtual touch tissue quantification values of the benign lesions (2.01±0.95m/s) differed from that of the malignant lesions (4.61±2.56m/s; P <0.001). The cutoff level for virtual touch tissue quantification value for malignancy was estimated to be 2.595m/s. Using the receiver operating characteristic curve curves with the cutoff value, the virtual touch tissue quantification value predicted malignancy with a sensitivity of 82.9%, specificity of 93.1% and gave an areas under the curve of 0.906 (95% CI 0.857–0.954). Acoustic radiation force impulse is feasible for cervical lymph nodes and provides quantitative elasticity measurements, which may complement B-mode ultrasound and potentially improve the characterization of cervical lymph nodes.

The purpose of this study was to test the morphology and haemodynamics of the renal artery in the rabbit as evaluated by conventional and contrast-enhanced ultrasonography (CEUS). The morphology and haemodynamics of the rabbit renal artery, including the diameter, which were measured using B-mode ultrasonography (US), colour Doppler US and CEUS, and systolic velocity, diastolic velocity and resistive index (RI) were measured using pulsed wave Doppler US. CEUS was used to measure the renal artery diameter: 0.21 ± 0.04 cm (right) and 0.21 ± 0.03 cm (left). Values of the main renal artery diameter obtained from CEUS significantly correlated with those of digital subtraction angiography. The blood flow velocity of the right main renal artery was 44.20 ± 8.71/18.92 ± 6.26 cm/s (systolic/diastolic) and 36.30 ± 6.89/17.64 ± 5.58 cm/s (systolic/diastolic), at its origin from the aorta and at the renal hilus, respectively. The blood flow velocity of the left main renal artery was 45.10 ± 8.49/19.00 ± 6.80 cm/s (systolic/diastolic) and 41.70 ± 10.25/19.55 ± 7.90 cm/s (systolic/diastolic), at its origin from the aorta and at the renal hilus, respectively. Conventional US provides a more feasible modality for measuring the morphology and haemodynamics of the rabbit renal artery. CEUS is a more accurate method for measuring diameter. This information on the morphology and haemodynamics of the rabbit renal artery might be helpful for researchers.

Objectives-Initial data suggest that elastography can improve the specificity of sonography for differentiating benign and malignant thyroid lesions. The primary objective of this study was to compare quantitative sonoelastography to conventional qualitative sonoelastography and sonography for thyroid nodule characterization.Methods-Ninety-eight thyroid masses (53 benign and 45 malignant) were examined with conventional sonography and sonoelastography. The images were classified into 4 patterns according to a previously proposed classification. In addition, strain ratios of thyroid tissue to the nodule were calculated. Receiver operating characteristic curve analysis was used to compare the diagnostic performance of the strain ratio and that of conventional sonography. The final diagnosis was obtained from histologic findings.Results-When a cutoff point of 3.79 was introduced, significantly different strain ratios for benign (mean +/- SD, 2.97 +/- 4.35) and malignant (11.59 +/- 10.32) lesions was obtained (P < .0001). The strain ratio measurement had 97.8% sensitivity and 85.7% specificity. The area under the curve for the strain ratio was 0.92, whereas that for the 4-point scoring system was 0.85. Of the conventional sonographic patterns, microcalcification had the highest area under the curve, at 0.72.Conclusions-Strain ratio measurement of thyroid lesions is a fast standardized method for analyzing stiffness inside examined areas. Used as an additional tool with B-mode sonography, it helps increase the diagnostic performance of the examination.