Cell-Graphs for Tissue Representation
and Quantification
Dr. Bulent Yener
Department of Computer Science
Rensselaer
Polytechnic Institute
Date:
Time:
Location: 367 Votey
Abstract
This talk introduces a novel mathematical technique called the cell-graphs to represent and quantify the information encoded in a tissue sample. The cell-graphs are generated from the low-cost, low-magnification tissue images according to the spatial distribution of the cells. The mathematical properties of cell-graphs provide a set of features that can be used by machine learning techniques to characterize the properties of the underlying tissue. It is shown that cell-graph approach leads to successful tissue diagnosis of a brain cancer called "malignant glioma". The cell-graph approach can also be used decision support system in several other areas including tissue engineering, measurement of drug responsiveness, and other types of cancer diagnosis and prognostication. In this talk we will first present the techniques to obtain cell-graphs of glioma and discuss future work.
This is joint work with Cigdem Demir at RPI and Hume Gultekin at OHSU.
About the speaker:
Bulent
Yener is an Associate Professor in the Department of Computer Science and Co-Director
of Pervasive Computing and
(This seminar is
hosted by Computer Science Student
Association.)