[Skip to Navigation]
November 1994

Discriminant Analysis of Alzheimer's Disease

Author Affiliations

Laboratory of Neurosciences National Institute on Aging National Institutes of Health Bethesda, MD 20892

Arch Neurol. 1994;51(11):1088. doi:10.1001/archneur.1994.00540230026006

We read the article by Killiany et al1 with considerable interest. We note that it confirmed our previously published data regarding the use of discriminant analysis to identify patients with early Alzheimer's disease (AD).2 However, the very preliminary nature of their observations deserves comment.

To date, we have studied 31 patients with dementia of the Alzheimer's type and have compared these with 29 healthy age- and sex-matched control subjects. All patients with dementia of the Alzheimer's type and control subjects were followed up longitudinally, including autopsy in some patients with dementia. This enabled us to test the diagnostic advantage of quantitative magnetic resonance imaging at the time of initial evaluation of patients subsequently confirmed to have probable AD.

Image segmentation3 determined various brain and cerebrospinal fluid volumes. Initial discriminant analysis revealed a significant gender difference among the control subjects. Subsequent discriminant analyses were applied according to gender.

Add or change institution