您目前所在位置: 新闻与学术交流   >   学术报告
美国阿肯色大学JingxianWu教授学术报告
发布时间:2018-06-19 09:01:08.0   阅读次数:568

西南交通大学创源大讲堂

Innovative Source Lecture Hall, Southwest Jiaotong University

无线通信与信息编码前沿系列讲座

Lecture Series: Frontiers in Wireless Communications & Information Coding

报告题目(Title): Unsupervised Bayesian Learning of Breast Cancer Detection with THz Imaging

报告专家(Speaker): Prof. Jingxian Wu, University of Arkansas, USA

报告时间(Time)2018620日,周下午2:30-3:30pm

报告地点(Venue)西南交通大学犀浦校区9号楼X9521#会议室

主持人(Chair): 范平志Pingzhi Fan)

内容提要Outline of the Talk:

Unsupervised Bayesian Learning of Breast Cancer Detection with THz Imaging

This talk presents an unsupervised Bayesian learning algorithm for cancer detection in Terahertz (THz) imaging of freshly excised murine tumors. Unlike most existing works with deterministic detection methods, we adopt a probabilistic learning approach that can iteratively calculate the probability each pixel in a THz image belonging to different types of tissues, such as cancer, fat, muscle, fibrous tissue, etc. Such a probabilistic approach produces important reliability information about the detection results that are not available in conventional methods. Specifically, under a Bayesian framework, a finite mixture model is used to represent the probability distributions of the intensities of pixels in the THz image, with each component in the mixture model corresponding to one tissue type. The prevalence of a specific type of tissue in a pixel can be represented through the weights of corresponding component to be learned through the data, without the need of labeled training data. Two different mixture models, Gaussian mixture and t-mixture models, are employed in the analysis. The empirical posterior distributions of parameters from both models are estimated by using a Markov chain Monte Carlo (MCMC) technique with Gibbs sampling. The performance of the algorithms is evaluated by comparing the detection results to their corresponding pathology results, and experiment results demonstrate the proposed algorithm can classify different tissue types with high accuracy. Overall, THz imaging shows good qualitative comparison to pathology.

报告人简介(Short Biography of the Speaker):

Jingxian Wu, University of Arkansas, USA ( http://comp.uark.edu/~wuj/ )

Jingxian Wu received the B.S. (EE) degree from the Beijing University of Aeronautics and Astronautics, Beijing, China, in 1998, the M.S. (EE) degree from Tsinghua University, Beijing, China, in 2001, and the Ph.D. (EE) degree from the University of Missouri at Columbia, Missouri, USA, in 2005. He is currently an Associate Professor with the Department of Electrical Engineering, University of Arkansas, Fayetteville. His research interests mainly focus on statistical signal processing, large scale data analytics, biomedical signal processing, and wireless communications. He served as symposium or track co-chairs for a number of international conferences, such as the 2012 IEEE International Conference on Communications, the 2009, 2015, and 2017 IEEE Global Telecommunications Conference, the 2017 International Conference on Communications in China, and the 2017 Wireless Communication and Signal Processing Conference, etc. He served as an Associate Editor of the IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY from 2007 to 2011 an Editor of the IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS from 2011 to 2016, and is now serving as an Associate Editor of the IEEE ACCESS.