University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Graduation Talks

title

Multivariate Analysis Applied to Image Data of Dementia Patients

speaker

Elisabeth Stühler, University Konstanz
Konstanz, Germany

date & place

Wednesday, 25.01.2012, 16:15 h
Room C252

abstract

Multivariate statistical analysis has recently proved to be superior to voxel-wise approaches when applied to molec- ular images. Our objective is to adapt and develop multivariate methods to advance the early and automated detection of various types of dementia based on SPECT and PET datasets. Main challenges of the project are the high dimensionality of the data and to ensure adequate robustness. First results show the feasibility of an adapted multiple discriminant analysis to discriminate more than two classes within a set of SPECT images. The main focus of future projects lies on automated detection of abnormal cases and on the correlation of molecular images with further biomarkers obtained by clinical examinations or extracted from MR images.