报告时间： 星期四（2016.3.31）2：30am
报告地点： 电子信息楼306
报告人：Ulas Bagci, Ph.D., Tenure-track Assistant Professor, University of Central Florida
邀请人： 陈新建 特聘教授
Abstract:
Separating Visceral Adipose Tissue (VAT) from Subcutaneous Adipose Tissue (SAT) is important in quantification of central obesity. We propose an unsupervised method combination of one-shot learning for body region detection, 3D CRF for fat segmentation, and local outlier detection for smoothing the results from CT scans. Meanwhile, we develop a method for automatically detecting and quantification brown adipose tissue (Brown fat, or BAT in short) from PET/CT images. Our proposed method is based on machine learning classifiers to detect potential regions first, then pet/ct co-segmentation algorithm based on a random walk, followed by a new probabilistic distance measure for differentiating brown fat from white fat. On more than 150 CT scans, we have obtained the state-of-the-art segmentation performance for SAT and VAT determination. The proposed CAD system for brown fat achieved a highly accurate region detection rate of more than 90%. Furthermore, BAT segmentation was achieved with more than 90% dice similarity score. The proposed method offers whole body adiposity analysis with white and brown fat, quantifying central obesity readily in seconds, and finding potential BAT regions to determine the metabolic health of the subjects. Future works will investigate the role of MRI for potential replacement to CT.
Biography:
Prof. Bagci is a faculty member at the Center for Research in Computer Vision (CRCV)<http://www.cs.ucf.edu/main.php>;, and the assistant professor in University of Central Florida (UCF). He is conducting research in biomedical imaginng, computer vision, clinical image processing, and statistical machine learning fields. Prior to CRCV, Prof. Bagci was a staff scientist and the lab manager at the NIH's Center for Infectious Disease Imaging (CIDI) Lab, department of Radiology and Imaging Sciences (RAD&IS). At NIH, he has developed and implemented educational and scientific research initiatives, and mentored postdoctoral and postbaccalaureate fellows for quantitative image analysis in infectious disease. Prof. Bagci had been the leading image analyst in biosafety/bioterrorism project initiated by NIAID and IRF. He obtained his PhD degree from School of Computer Science, University of Nottingham (UK) in collaboration with Radiology department of University of Pennsylvania (with Prof. Udupa, MIPG). He has masters from Electrical Engineering and Computer Sciences and certificates of mastery from Harvard, Berkeley, MIT, and Johns Hopkins Universities in the topics of statistics, public health, and clinical trials.
Prof. Bagci is a faculty member at the Center for Research in Computer Vision (CRCV)<http://www.cs.ucf.edu/main.php>;, and the assistant professor in University of Central Florida (UCF). He is conducting research in biomedical imaginng, computer vision, clinical image processing, and statistical machine learning fields. Prior to CRCV, Prof. Bagci was a staff scientist and the lab manager at the NIH's Center for Infectious Disease Imaging (CIDI) Lab, department of Radiology and Imaging Sciences (RAD&IS). At NIH, he has developed and implemented educational and scientific research initiatives, and mentored postdoctoral and postbaccalaureate fellows for quantitative image analysis in infectious disease. Prof. Bagci had been the leading image analyst in biosafety/bioterrorism project initiated by NIAID and IRF. He obtained his PhD degree from School of Computer Science, University of Nottingham (UK) in collaboration with Radiology department of University of Pennsylvania (with Prof. Udupa, MIPG). He has masters from Electrical Engineering and Computer Sciences and certificates of mastery from Harvard, Berkeley, MIT, and Johns Hopkins Universities in the topics of statistics, public health, and clinical trials.