报告时间：2019.10.21周一上午10:00

地点：信息科学与工程学院220（原106）

报告人: Binsheng Zhao, D.Sc.,Director, Computational Image Analysis Laboratory

       Professor, Columbia University Medical Center, New York, New York, USA

       Website: https://www.columbiaradiology.org/cialab

报告简介:

Quantitative imaging biomarkers (QIBs) play increasingly important roles in the era of precision medicine. Radiomics, a type of QIBs, refers to the comprehensive quantification of tumor phenotypes that may be linked to genotypes/clinical outcomes by analyzing high dimensional quantitative features extracted from non-invasive routine radiographic images. A growing body of literature has shown the promise of radiomics features and models in supporting clinical decision-making in cancer detection, diagnosis, prognosis, and response prediction and assessment with its added value. However, a qualified QIB must be practically obtainable with efficient and accurate software/tools, reproducible and robust across heterogeneous imaging acquisition settings, and correlated with genotypes and/or clinical outcomes. This lecture will discuss these challenges as well as some future research directions.

报告人简介:

Binsheng Zhao graduated from National University of Defense Technology, Changsha, with B.Sc. (1984) and M.Sc. (1987) in Electrical Engineering. In 1994, she received D.Sc. in Medical Informatics from University of Heidelberg, Germany. Dr. Zhao has worked for 20+ years at the interface of medical physics, radiology and oncology. Since early 2002, she has led a research team to develop computer-aided tumor/organ/tissue segmentation and characterization methods using CT and MRI. Her goal throughout has been to investigate new quantitative imaging biomarkers for better cancer diagnosis, response prediction and assessment and to automate and optimize these quantitative imaging biomarkers; recently, her interest has naturally led to a deep engagement with radiomics and deep machine learning. Since 2004, Dr. Zhao has been playing important roles (e.g., as steering committee member and committee member) in National Cancer Institute (NCI) and Radiological Society of North America (RSNA) initiated quantitative imaging biomarker projects that strive to aid development and validation of quantitative imaging biomarkers for improved cancer detection, diagnosis, and treatment. In 2009, her team published an unprecedented work, known as same-day repeat CT study, describing their findings of the minimally detectable tumor change by CT. The findings are essential to re-evaluating conventional response assessment methods and to establishing new volumetric response criteria.

The repeat CT dataset has been made publicly available through NCI for researchers worldwide,resulting in numerous additional publications. Dr. Zhao’s subsequent clinical study correlating early radiographic changes with EGFR mutation status in non-small cell lung cancer treated with targeted therapy, demonstrated, for the first time, improved correlations by the volumetric technique over the conventional diameter method. Her publication “Reproducibility of radiomics for deciphering tumor phenotype with imaging” kicked off recent intense discussions on the importance of developing reproducible and robust quantitative imaging biomarkers across heterogeneous imaging acquisition settings. Dr. Zhao served / is serving as PI or key investigator in numerous NIH studies. She has published over 100 peer-reviewed articles and holds 4 patents, and has successfully translated some of her team-developed algorithms from bench to bedside.