General Departmental Seminar Series
Surface Data Smoothing and Analysis in the Autistic Brain
Moo Chung, PhD, Dept. of Biostatistics and Medical Informatics, UW-Madison
Friday, September 17, 2004, 12-1 p.m.
G5/113 Clinical Sciences Center, 600 Highland Ave.
The advancement in brain imaging has produced abundant brain surface data. However, it is not so straightforward to analyze such data due to curved geometry. Gaussian kernel smoothing has been widely used in 3D medical images as a way to increase signal-to-noise ratio and smoothness of dependent noise structure, in part, due to its simplicity in numerical implementation. Gaussian kernel is isotropic in Euclidian space so it assigns the same weights to observations equal distance apart. However when we smooth data residing on the brain cortex, it fails to be isotropic. On the curved surface, a straight line between two points is not the shortest distance so one may incorrectly assign less weights to closer observations. Then the question is how to correctly formulate isotropic kernel smoothing on the cortex. We have recently developed a new technique called heat kernel smoothing which addresses this problem with some interesting properties. The concept of heat kernel smoothing will be introduced. As an application we detect the regions of abnormal cortical gray matter thickness in a group of autistic subjects via random field based multiple comparison correction that utilizes the new smoothing technique.
Return to seminar list