Geometric modeling and statistical analysis of anatomical trees

Speaker: Aasa Feragen , University of Copenhagen

Date: Wednesday, November 06, 2013

Time: 4:00 PM to 5:00 PM Note: all times are in the Eastern Time Zone

Public: Yes

Location: 32-D507

Event Type:

Room Description:

Host: Polina Golland

Contact: Polina Golland, x38005, polina@csail.mit.edu

Relevant URL:

Speaker URL: None

Speaker Photo:
None

Reminders to:

Reminder Subject: TALK: Geometric modeling and statistical analysis of anatomical trees

Trees appear frequently in anatomy, for instance as transportation
systems to distribute blood, water or air. Being able to do statistics
on populations of trees is essential to understanding disease and
development, due to the critical role of transportation systems in the
organs they are part of. This is made difficult by the anatomical
variation in branching structure across subjects. While a multitude of
ad hoc methods exist for solving discrete-valued problems such as
classification, there are few well-founded tools available for solving
tree-valued problems, which are particularly important for visualizing
and understanding the data.

I will discuss a mathematical framework for analyzing trees whose
branches have properties such as branch length or shape, described by
an element of a vector space. Such trees will be called 'geometric
trees'. I will discuss a tree-shape space model for geometric trees
with a fixed set of leaf labels, and its application to several
problems relating to the analysis of airway trees from human lungs.
Moreover, I will discuss how this model relates to more general
approaches to tree- and network analysis, and discuss a number of open
problems related to these more general models.

The presented work is carried out in collaboration with Megan Owen,
Francois Lauze, Jens Petersen, Mads Nielsen, Asger Dirksen and Marleen
de Bruijne.

Research Areas:

Impact Areas:

See other events that are part of the Biomedical Imaging and Analysis 2013/2014.

Created by Polina Golland Email at Thursday, October 17, 2013 at 11:29 AM.