Geometric Statistically Based Methods For Medical Image Computing
发布日期：2013-03-10 16:18:04 浏览次数：次
Time：10:00 am, March. 21 (Thursday)
Location: 104#, CBIR
Speaker: Yi Gao (Harvard Medical School)
Yi is a Research Fellow at the Harvard Medical School. He joined the Psychiatry Neuroimaging Laboratory in 2011. Before that, he received his Ph.D. in Biomedical Engineering in November 2010 from the Georgia Institute of Technology. He obtained a Masters in Mathematics also from the Georgia Institute of Technology and a Masters and Bachelors in Biomedical Engineering from the Tsinghua University, Beijing, China in 2005 and 2003. He has also worked as an algorithm development intern in General Electric Healthcare and Siemens Corporate Research in 2005 and 2007, respectively. His research interests lie in the areas of image analysis, shape analysis, and computer vision.
Geometric statistically based methods for medical image computing
In this talk, we report some of our researches in the medical image computing, for both diagnosis and intervention purposes. The methods presented here include segmentation, registration, shape analysis, and longitudinal study. In addition to algorithm development, we also emphasis here the realization of the algorithm into publically available software platforms so that the algorithm will be readily delivered to the clinical end-users for their evaluation and for translation.
The algorithms presented here are mostly naturally based on the geometry and statistical principles. Specifically, the statistical learning, multi-scale analysis, Riemann-Finsler Geometry, and Lie Group/Lie Algebra theory are synergized into common frameworks for the purposes that include: automatic identification of critical organs and structures; registration of three-dimensional geometric objects; and assessing the longitudinal shape differences among populations.