Meeting Banner
Abstract #0971

Multivariate Analysis of Diffusion Tensor Metrics in Mild Cognitive Impairment & Healthy Aging

Yu Zhang1, 2, Norbert Schuff1, 2, Kristine Yaffe2, Howard Rosen2, Bruce Miller2, Michael Weiner1, 2

1Center for Imaging of Neurodegenerative Diseases, VA Medical Center, San Francisco, San Francisco, CA, United States; 2University of California, San Francisco, San Francisco, CA, United States

Previous studies have used univariate tests of diffusion tensor metrics such as fractional anisotropy (FA) or radial diffusivity (DR) to classify mild cognitive impairment (MCI) subjects and healthy elderly controls. This study applied multivariate tests of DTI, including simultaneously all three diffusion tensor eigenvalues in a large sample of 54 MCI and 66 control subjects. The results show that multivariate tests of the diffusion eigenvalues detect regions of white matter alterations more consistently than univariate-tests. This method has potential to identify early cognitive impairment.

Abstract PDF Poster PDF

Keywords

acceleration acquisition adjusted aging aligned alterations analyses anisotropy anterior appeared applied association atlas bias bilateral body capture captured caudal cognitive concept considered consistent consistently continuous contrast control controls corpus corrected covariance cross deep detect detecting detection detects determine diagnosis diffusion diffusivity discovery disease diseases distortions distribution distributions early eddy eigenvalues either elderly entirely excluded false fiber fibers fold fractional gender generally goal healthy identify impacted impairment imported included individual isthmus labeled lesions limbic linear lists measured medical metrics mild miller mixed models multivariate noise onto package parallel participated posterior potential precursor predicted previous radial radiology reduce reduction regional reported resolution sensitive sensitivity sensitizing sets sided significance slices space specifically statistics stria studies subjects susceptibility system table tensor tool transform univariate usually visual whereas white