Transcript MTNS - Medical image computing

MINERVA GROUP @ Georgia Tech
People involved with NAMIC
 Professor Allen Tannenbaum
 Students:






Ramsey Al-Hakim
Jimi Malcolm
John Melonakos
Delphine Nain
Eric Pichon
Yogesh Rathi
http://www.bme.gatech.edu/groups/bil/
1
Research Topics in our Group
Topics relevant to NAMIC:
 PDE’s for image processing
 Variational and Statistical methods for Segmentation
and Registration
 Shape analysis
 Stochastic Curve/Surface Evolution
2
Year 1: Segmentation
Statistical Region Growing (Eric Pichon, in
Slicer “FastMarching” Module)
 Unidirectional evolution allows for fast implementation
(“Fast Marching”)
 Principled general purpose approach. Use Parzen
windows to estimate probability density function.
(Using non-parametric statistics means no assumption
on data)
Real MRI, comparison
with manual segmentations
(Surgical Planning Lab)
Eric Pichon, Allen Tannenbaum, and Ron Kikinis. A statistically based flow for image segmentation.
Medical Image Analysis, 8(3):267-274, September 32004
Year 1: Image Smoothing
Image Smooth (Yogesh Rathi, in Slicer)
 2D and 3D smoothing of images performed using the
geometric heat equation, where level lines of the
image are smoothed according to their curvature
(kappa).
 Kappa raised to the 1/3 performs smoothing for each
of the slices, 'slice-by-slice'.
 Kappa raised to the 1/4 performs smoothing in the zdirection as well, hence it is more accurate.
4
Years 1&2
Shape Analysis (Delphine Nain)
Statistical Segmentation & Registration (John
Melonakos, Ramsey Al-Hakim, Jimi Malcolm)
5