Volume-based vs. voxel-based brain morphometry for Alzheimer’s disease prediction

Orateur : Alexis ROCHE (Siemens Biomedical Imaging Centre, CIBM, Lausanne)

Automated brain morphometry using conventional T1-weighted images has proved effective to quantify anatomical changes related to normal development and progression of various diseases. Brain morphometry methods include voxel-based morphometry as implemented for instance in SPM, and volume-based morphometry as in FreeSurfer or Siemens brain morphometry work-in-progress package, where a subject is represented by volumes of different anatomical structures. This work aims at comparing voxel-based and volume-based morphometry in the context of assistance to diagnosis of Alzheimer’s disease. After reviewing the basic algorithmic principles behind the different methods and discussing their respective pros and cons, I will present experimental results of automated disease prediction and patient classification from the standardized Alzheimer’s Disease Neuroimaging Initiative (ADNI) data sets.

Télécharger dans mon agenda