BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Learning distributions of shape trajectories: a hierarchical model
  on a manifold of diffeomorphisms - Alexandre Bône (INSERM\; INRIA\; Univ
 ersité Pierre et Marie Curie Paris)
DTSTART:20171116T110000Z
DTEND:20171116T113000Z
UID:TALK95146@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:<span>Co-authors: Olivier Colliot		(CNRS)\, Stanley Durrleman	
 	(INRIA)        <br></span><span><br>We propose a mixed effects statistica
 l model to learn a distribution of shape trajectories from longitudinal da
 ta\, i.e. the collection of individual objects repeatedly observed at mult
 iple time-points. Shape trajectories and their variations are defined via 
 the action of a group of deformations. The model is built on a generic sta
 tistical model for manifold-valued longitudinal data\, for which we propos
 e to use a finite-dimensional set of diffeomorphisms with a manifold struc
 ture\, an efficient numerical scheme to compute parallel transport on this
  manifold and a specific sampling strategy for estimating shapes within a 
 Markov Chain Monte Carlo (MCMC) method.  The method allows the estimation 
 of an average spatiotemporal trajectory of shape changes at the group leve
 l\, and the individual variations of this trajectory in terms of shape and
  pace of shape changes. This estimation is obtained by a Stochastic Approx
 imation of the Expectation-Maximization (MCMC-SAEM). We show that the algo
 rithm recovers the optimal model parameters with simulated 2D shapes. We a
 pply the method to estimate a scenario of alteration of the shape of the h
 ippocampus 3D brain structure during the course of Alzheimer&#39\;s diseas
 e.</span>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR