Search

Your search keyword '"Deng, Wenxiang"' showing total 221 results
Start Over Author "Deng, Wenxiang"
221 results on '"Deng, Wenxiang"'

9. Unsupervised anomaly localization using VAE and beta-VAE

Author

Zhou, Leixin, Deng, Wenxiang, and Wu, Xiaodong

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning
Abstract

Variational Auto-Encoders (VAEs) have shown great potential in the unsupervised learning of data distributions. An VAE trained on normal images is expected to only be able to reconstruct normal images, allowing the localization of anomalous pixels in an image via manipulating information within the VAE ELBO loss. The ELBO consists of KL divergence loss (image-wise) and reconstruction loss (pixel-wise). It is natural and straightforward to use the later as the predictor. However, usually local anomaly added to a normal image can deteriorate the whole reconstructed image, causing segmentation using only naive pixel errors not accurate. Energy based projection was proposed to increase the reconstruction accuracy of normal regions/pixels, which achieved the state-of-the-art localization accuracy on simple natural images. Another possible predictors are ELBO and its components gradients with respect to each pixels. Previous work claimed that KL gradient is a robust predictor. In this paper, we argue that the energy based projection in medical imaging is not as useful as on natural images. Moreover, we observe that the robustness of KL gradient predictor totally depends on the setting of the VAE and dataset. We also explored the effect of the weight of KL loss within beta-VAE and predictor ensemble in anomaly localization., Comment: arXiv admin note: substantial text overlap with arXiv:2002.03734 by other authors

Published
2020

13. Robust Image Segmentation Quality Assessment

Author

Zhou, Leixin, Deng, Wenxiang, and Wu, Xiaodong

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Deep learning based image segmentation methods have achieved great success, even having human-level accuracy in some applications. However, due to the black box nature of deep learning, the best method may fail in some situations. Thus predicting segmentation quality without ground truth would be very crucial especially in clinical practice. Recently, people proposed to train neural networks to estimate the quality score by regression. Although it can achieve promising prediction accuracy, the network suffers robustness problem, e.g. it is vulnerable to adversarial attacks. In this paper, we propose to alleviate this problem by utilizing the difference between the input image and the reconstructed image, which is conditioned on the segmentation to be assessed, to lower the chance to overfit to the undesired image features from the original input image, and thus to increase the robustness. Results on ACDC17 dataset demonstrated our method is promising.

Published
2019

18. Traditional Chinese medicine Master XIONG Jibo’s medication experience in treating arthralgia syndrome through data mining

Author

Deng Wenxiang, Zhang Jidong, Zhang Wenan, and He Qinghu

Subjects
Traditional Chinese medicine Master, XIONG Jibo, Arthralgia syndrome, Data mining, Natural language processing (NLP), medication experience, Medicine, Other systems of medicine, RZ201-999
Abstract

Objective: This study aimed to examine and propagate the medication experience and group formula of traditional Chinese medicine (TCM) Master XIONG Jibo in diagnosing and treating arthralgia syndrome (AS) through data mining. Methods: Data of outpatient cases of Professor XIONG Jibo were collected from January 1, 2014 to December 31, 2018, along with cases recorded in A Real Famous Traditional Chinese Medicine Doctor: XIONG Jibo's Clinical Medical Record 1, which was published in December 2019. The five variables collected from the patients’ data were TCM diagnostic information, TCM and western medicine diagnoses, syndrome, treatment, and prescription. A database was established for the collected data with Excel. Using the Python environment, a customized modified natural language processing (NLP) model for the diagnosis and treatment of AS by Professor XIONG Jibo was established to preprocess the data and to analyze the word cloud. Frequency analysis, association rule analysis, cluster analysis, and visual analysis of AS cases were performed based on the Traditional Chinese Medicine Inheritance Computing Platform (V3.0) and RStudio (V4.0.3). Results: A total of 610 medical records of Professor XIONG Jibo were collected from the case database. A total of 103 medical records were included after data screening criteria, which comprised 187 times (45 kinds) of prescriptions and 1 506 times (125 kinds) of Chinese herbs. The main related meridians were the liver, spleen, and kidney meridians. The properties of Chinese herbs used most were mainly warm, flat, and cold, while the flavors of herbs were mainly bitter, pungent, and sweet. The main patterns of AS included the damp heat, phlegm stasis, and neck arthralgia. The most commonly used herbs for AS were Chuanniuxi (Cyathulae Radix), Huangbo (Phellodendri Chinensis Cortex), Cangzhu (Atractylodis Rhizoma), Qinjiao (Gentianae Macrophyllae Radix), Gancao (Glycyrrhizae Radix et Rhizoma), Huangqi (Astragali Radix), and Chuanxiong (Chuanxiong Rhizoma). The most common effect of the herbs was “promoting blood circulation and removing blood stasis”, followed by “supplementing deficiency (Qi supplementing, blood supplementing, and Yang supplementing)”, and “dispelling wind and dampness”. The data were analyzed with the support ≥ 15% and confidence = 100%, and after de-duplication, five second-order association rules, 39 third-order association rules, 39 fourth-order association rules, and two fifth-order association rules were identified. The top-ranking association rules of each were “Cangzhu (Atractylodis Rhizoma) → Huangbo (Phellodendri Chinensis Cortex)” “Cangzhu (Atractylodis Rhizoma) + Chuanniuxi (Cyathulae Radix) → Huangbo (Phellodendri Chinensis Cortex)” “Chuanniuxi (Cyathulae Radix) + Danggui (Angelicae Sinensis Radix) + Gancao (Glycyrrhizae Radix et Rhizoma) → Qinjiao (Gentianae Macrophyllae Radix)” and “Chuanniuxi (Cyathulae Radix) + Danggui (Angelicae Sinensis Radix) +Gancao (Glycyrrhizae Radix et Rhizoma) + Huangbo (Phellodendri Chinensis Cortex) → Qinjiao (Gentianae Macrophyllae Radix)”, respectively. Five clusters were obtained using cluster analysis of the top 30 herbs. The herbs were mainly drying dampness, supplementing Qi, and promoting blood circulation. The main prescriptions of AS were Ermiao San (二妙散), Gegen Jianghuang San (葛根姜黄散), and Huangqi Chongteng Yin (黄芪虫藤饮). The herbs of core prescription included Cangzhu (Atractylodis Rhizoma), Chuanniuxi (Cyathulae Radix), Gancao (Glycyrrhizae Radix et Rhizoma), Huangbo (Phellodendri Chinensis Cortex), Mugua (Chaenomelis Fructus), Qinjiao (Gentianae Macrophyllae Radix), Danggui (Angelicae Sinensis Radix), and Yiyiren (Coicis Semen). Conclusion: Clearing heat and dampness, relieving collaterals and pain, and invigorating Qi and blood are the most commonly used therapies for the treatment of AS by Professor XIONG Jibo. Additionally, customized NLP model could improve the efficiency of data mining in TCM.

Published
2022
Full Text
View/download PDF

31. Deep-Learning-Based Estimation of 3D Optic-Nerve-Head Shape from 2D Color Fundus Photographs in Cases of Optic Disc Swelling

Author

Islam, Mohammad Shafkat, Wang, Jui-Kai, Deng, Wenxiang, Thurtell, Matthew J., Kardon, Randy H., Garvin, Mona K., Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Woeginger, Gerhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Fu, Huazhu, editor, Garvin, Mona K., editor, MacGillivray, Tom, editor, Xu, Yanwu, editor, and Zheng, Yalin, editor

Published
2020
Full Text
View/download PDF

35. Region-Based Segmentation of Capillary Density in Optical Coherence Tomography Angiography

Author

Deng, Wenxiang, Tamplin, Michelle R., Grumbach, Isabella M., Kardon, Randy H., Garvin, Mona K., Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Woeginger, Gerhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Fu, Huazhu, editor, Garvin, Mona K., editor, MacGillivray, Tom, editor, Xu, Yanwu, editor, and Zheng, Yalin, editor

Published
2019
Full Text
View/download PDF

38. Event‐triggered adaptive estimated inverse neural network control of uncertain nonlinear systems with unknown hysteresis effects.

Author

Zhou, Ning, Yao, Jianyong, and Deng, Wenxiang

Subjects
NONLINEAR systems, RADIAL basis functions, HYSTERESIS, UNCERTAIN systems
Abstract

This article investigates an event‐triggered adaptive estimated inverse control scheme for uncertain nonlinear systems with hysteresis effects, parametric uncertainties, and unknown disturbances. An online estimated inverse hysteresis compensation mechanism is developed, in which an adaptive technique is designed to obtain the value of unknown hysteresis parameters. Compared with common approaches, its biggest advantage lies in the fact that it eliminates the need for experimental determination of hysteresis parameters, thereby reducing time‐consuming offline identification work and enhance the compatibility of the proposed method. Additionally, an adaptive radial basis functions neural network is applied to approximate the unknown disturbances, and its weight coefficients, along with unknown system parameters, are estimated by means of the adaptive method. Furthermore, the introduction of relative threshold event‐triggered control significantly reduces communication costs resulting from the hysteresis compensation. Through Lyapunov analysis, the proposed controller guarantees all signals are bounded and the errors are convergent. Numerical simulation results demonstrate the superiority of the developed controller. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results