YWYu Wang

Yu Wang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Yu Wang has a total of 25 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Measurement of Echocardiographic Parameters in Various Cardiac Phases: A Comparative Study of Deep Learning Models and Keypoint Representations

    by Ruyi Zhang, Aijiao Sun, Danyang Zhang, Dingcheng Tian, Yudong Yao, Yu Wang, Feifei Sun

    The assessment of the parasternal long-axis (PLAX) view is a critical task during echocardiographic examination. Accurate measurement and quantification of anatomical structures in PLAX view is essential for the diagnosis, treatment, and prognosis of cardiac diseases. Yet, manual measurement is limited by its time-consuming nature and susceptibility to interindividual variability and image quality fluctuations. Automated keypoint detection of anatomical structures using deep learning offers a robust solution, potentially assisting physicians in the precise measurement of parameters within specified views. A systematic comparison of various deep learning models and keypoint representations is conducted in this article for keypoint detection within the PLAX view, with the aim of exploring the best models and representations. Three datasets, comprising 442 patients and including data from end-diastole (ED), mid-systole (MS), and end-systole (ES), are constructed to fully evaluate the accuracy and generalizability of different deep learning methods. In the 12-keypoint detection task conducted on the ED dataset, the HRNet-heatmap method achieved the highest success detection rate (SDR) of 68.77% and the lowest mean absolute error (MAE) of 3.03 mm among the evaluated approaches. For the 2-keypoint detection task on the MS dataset, the HRFormer-heatmap method performed the best, with an SDR of 89.09% and an MAE of 1.90 mm. In the 2-keypoint detection task on the ES dataset, the CSPNeXt-SimCC method demonstrated the best performance, achieving an SDR of 94.46% and an MAE of 2.95 mm.


  • Open Access

    CT Lung Nodule Segmentation: A Comparative Study of Data Preprocessing and Deep Learning Models

    by Weihao Chen, Yu Wang, Dingcheng Tian, Yudong Yao

    The number of deaths from lung cancer reached 1.8 million in 2020, ranking first among all cancers. Early diagnosis has been found to improve the survival rate of lung cancer patients after treatment in clinical care. Computed tomography (CT) is a technique commonly used in the early detection of lung cancer to determine the benignity or malignancy of lung nodules. Manual analysis of CT results is less efficient and its accuracy is affected by physicians’ experience levels. Segmenting lung nodules in a computer-aided diagnosis (CAD) system can effectively improve the efficiency and accuracy of the diagnosis. In this paper, we evaluate several deep learning segmentation models (including UNet, SegNet, GCN, FCN, DeepLabV3+, PspNet TransUNet, SwinNet) and examine the effects of different preprocessing methods on the models to explore the best preprocessing and training strategies for lung nodule segmentation. Specifically, we investigate the effects of two different data preprocessing methods (adding a lung mask and croping the region of interest) on the segmentation results, where better segmentation results are achieved by including the nodal data of the region of interest without the lung mask. Through a comprehensive comparison, TransUNet achieves the best segmentation accuracy, with DICE indices of 0.887, 0.871, 0.75, and 0.744 tested on four datasets, respectively.


  • Open Access

    A comparative study on fetal head circumference measurement from ultrasound images using deep learning models

    by Yu Wang, Yudong Yao

    Ultrasound imaging is the most commonly used imaging modality for the prenatal examination of pregnant women, with real-time imaging and no radiation characteristics. Through a ultrasound image of the fetal head, doctors can measure the fetal head circumference (HC) to evaluate fetal growth and potential delivery mode. In practice, fetal HC is usually measured manually by doctors based on ultrasound images. Manual measurement of fetal HC is subjective and time-consuming, which has a negative impact on measurement accuracy and efficiency. At present, deep learning is widely investigated in the medical field. Many researchers apply deep learning to measuring fetal HC to assist doctors to accurately and quickly completing the measurement of fetal HC. In this paper, we compare the performance of eight deep learning models (U-Net, Attention U-Net, GINet, global reasoning unit (GloRe), SegFormer, Segmenter, BiSeNet V2, and short-term dense concatenate network (STDC)) on two fetal HC measurement datasets. SegFormer achieves the best results in Dice similarity coefficient (DSC), Hausdorff distance (HD), and absolute Difference (ADF). The performance of Attention U-Net is slightly worse than that of SegFormer.


  • Open Access

    Morphologic CT and MRI features of primary parotid squamous cell carcinoma and its predictive factors for differential diagnosis with mucoepidermoid carcinoma

    by Xiaohua Ban, Huijun Hu, Yue Li, Lingjie Yang, Yu Wang, Rong Zhang, Chuanmiao Xie, Cuiping Zhou, Xiaohui Duan

    Background: Primary parotid squamous cell carcinoma (SCC) is a rare entity with a poor prognosis. Pathologically, the diagnosis of it requires the exclusion of parotid mucoepidermoid carcinoma (MEC). Currently, the imaging features of primary parotid SCC and the predictive indicators for differential diagnosis of the two entities have not been well reported. Our purpose was to identify the imaging characteristics of primary parotid SCC and to determine the predictive factors for its’ differential diagnosis.Results: Thirty-one participants with primary parotid SCC and 59 with primary parotid MEC were enrolled. Clinical, CT and MRI features were reviewed and compared by univariate analysis. Then, multinomial logistic regression was used to determine the predictors to distinguish parotid SCC from MEC. Most primary parotid SCCs exhibited irregular shape, ill-defined margin, incomplete or no capsule, heterogeneous and marked or moderate enhancement, necrosis, local tumor invasiveness (LTI). Age, maximal dimension, shape, degree of enhancement, gradual enhancement, necrosis, and LTI were different between the primary parotid SCCs and MECs in univariate analysis (p < 0.05). While in multinomial logistic regression analysis, only age and necrosis were the independent predictors for distinguishing parotid SCC from MEC, and this model exhibited an area under curve of 0.914 in ROC curve analysis.Conclusions: Primary parotid SCC has some distinct imaging features including the large tumor size, irregular shape, ill-defined margin, and particularly the marked central necrosis. Patients with age ≥ 51.5 years and necrosis on the image of the primary tumor in the parotid gland could be more likely to be SCCs than MECs.


  • Top Cited Papers
    Open Access

    Deep Learning in Medical Ultrasound Image Analysis: A Review

    by Yu Wang, Xinke Ge, He Ma, Shouliang Qi, Guanjing Zhang, Yudong Yao

    Ultrasound (US) is one of the most widely used imaging modalities in medical diagnosis. It has the advantages of real-time, low cost, noninvasive nature, and easy to operate. However, it also has the unique disadvantages of strong artifacts and noise and high dependence on the experience of doctors. In order to overcome the shortcomings of ultrasound diagnosis and help doctor improve the accuracy and efficiency of diagnosis, many computer aided diagnosis (CAD) systems have been developed. In recent years, deep learning has achieved great success in computer vision with its unique advantages. In the aspect of medical US image analysis, deep learning has also been exploited for its great potential and more and more researchers apply it to CAD systems. In this paper, we first introduce the deep learning models commonly used in medical US image analysis; Second, we review the data preprocessing methods of medical US images, including data augmentation, denoising, and enhancement; Finally, we analyze the applications of deep learning in medical US imaging tasks (such as image classification, object detection, and image reconstruction).


  • Open Access

    Forecast Method of Energy Public Opinion Based on Conditional Random Field

    by Yu Wang, Shengzhen Xie, Junri Tang, Ling Sui, Kaiqi Kou, Zhiqiong Wang

    As the driving force of development in the new century, the energy problem is an important factor restricting economic development, which is related to a country’s economic security and national security. The public’s views or attitudes towards (heat events related to energy issues, such as the State Grid’s repeated public attention during the 2016 electricity reform) significantly affect the government’s monitoring and guidance of the oil crisis. Ordinary research on emotion prediction regards the change of public attitude towards events as a random process, only considering the influence of the attitude tendency held by users in the early stage on the later stage of emotion development, and not considering the influence of external factors on emotion prediction in the actual situation. This paper proposes a public opinion emotion prediction method based on conditional random fields. Taking the transition probability between external factors as the characteristic transition matrix of the conditional random field, considering the influence of external factors on the development of events, and combining with bidirectional LSTM, the BILSTM-CRF model in this paper is proposed. Experiments show that the emotion prediction method based on conditional random fields can predict the emotion state more accurately, which verifies the superiority of the emotion prediction method based on conditional random fields.