Regression Modeling on Manifolds with Application to Image Analysis

Friday, February 2, 2018 - 11:00am EST

  • Location:2 MetroTech Center, 10th Floor, 10.099
  • Contact:Guido Gerig
    gerig@nyu.edu

Speaker:  P. Thomas Fletcher, University of Utah

Abstract: 

Manifold representations of data derived from images, particularly representations of shapes and image transformations, have important applications in biology, medicine, and computer vision. In this talk, I will focus on recent work by my group developing regression models on shape manifolds. These models capture complex, nonlinear deformations between shapes and trajectories of shape change over time. I will show how classical linear and polynomial regression models for Euclidean data can be naturally extended to manifolds. This includes Bayesian formulations of regression, which automatically determine model complexity, and models of longitudinal data, i.e., repeated measures of the same individuals over time. I will show applications of this work in neuroimaging studies to understand both healthy aging and neurodegenerative diseases.


Bio: 

P. Thomas Fletcher is an Associate Professor in the School of Computing at the University of Utah  and  works within the Scientific Computing and Imaging Institute. His research focuses on solving problems in medical image analysis and computer vision through the combination of statistics and differential geometry.