Math, Data and Medical Imaging

Abstract:  In this overview talk, we review several practical problems in medical image analysis and various approaches to their solution. Directly comparing medical images involves image registration, also called alignment, which can be accomplished with the tools of differential geometry, machine learning, or a combination of these. We introduce some of these methods, and also a mathematically related problem in functional data analysis. Understanding medical images often involves their segmentation into anatomical parts, as well as the identification of non-standard “anomalies”
that may indicate disease. We review some progress in stroke lesion detection and introduce a general method for the detection of anomalies of unknown type, using Bayesian neural networks. Finally, we discuss the long-term goal of integrated probabilistic models for anatomy and physiology in
the context of precision medicine. 

Bio: Dr. Tanya Schmah earned a BScmHonours in Mathematics with a Major in Computer Science from the University of Western Australia. She worked for 5 years as a Computer Systems Analyst then earned an MA from Bryn Mawr College, US, followed by a PhD from the École Polytechnique Fédérale de Lausanne (Swiss Federal Institute of Technology) in 2001, in the field of geometric mechanics. After two years as a Lecturer at the University of Warwick, she joined the faculty of Macquarie University, Australia, earning tenure in 2005. In 2007 she joined the Machine Learning group at the University of Toronto as a postdoctoral fellow, working with Richard Zemel and Geoffrey Hinton as well as with Stephen Strother at the Rotman Research Institute at Baycrest. In 2015 she joined the Department of Mathematics and Statistics at the University of Ottawa, where she is currently an Associate Professor.