Yasuo Yamashita RT, Presenter: Nothing to Disclose

Hidetaka Arimura PhD, Abstract Co-Author: Nothing to Disclose

Takashi Yoshiura MD, PhD, Abstract Co-Author: Nothing to Disclose

Chiaki Tokunaga, Abstract Co-Author: Nothing to Disclose

Akira Monji MD, PhD, Abstract Co-Author: Nothing to Disclose

Fukai Toyofuku PhD, Abstract Co-Author: Nothing to Disclose

Jyunpei Kuwazuru BS, Abstract Co-Author: Nothing to Disclose

Taiki Magome, Abstract Co-Author: Nothing to Disclose

Yasuhiko Nakamura, Abstract Co-Author: Nothing to Disclose

Nobuyoshi Ohya, Abstract Co-Author: Nothing to Disclose

Hiroshi Honda MD, Abstract Co-Author: Nothing to Disclose

Masafumi Ohki PhD, Abstract Co-Author: Nothing to Disclose

Our preliminary results suggest that our proposed method based on functional image features obtained by the ASL technique would be feasible and may be clinically useful in detection of patients with AD.

Arterial spin labeling (ASL) is a non-invasive technique for measurement of cerebral blood flow (CBF), which would decrease in specific regions of patients with Alzheimer’s disease (AD). We have developed a computer-aided classification system for AD patients based on functional image features derived from the CBF maps measured by the ASL.

For evaluation of our proposed method, we applied it to magnetic resonance (MR) data of 15 patients with AD and 15 control subjects, which were scanned on a 3.0-T MR unit. The proposed method consisted of two steps. At the first step, the average CBFs in 16 cortical regions of frontal, limbic, occipital, parietal, sub-lobar, temporal lobes, posterior cingulate gyri and precunei were determined based on the CBF map images obtained by the ASL technique of MR imaging. The Talairach brain atlas, which is labeled for functional human brain mapping, was applied for extraction of the cortical regions. For that purpose, the registration between CBF map image and Talairach brain atlas was performed based on an affine transformation and a free-form deformation with a B-spline function. In the second step, the 30 cases were classified into AD and non-AD categories by using a support vector machine, which learned the CBF values in the following six regions, i.e., four regions (left occipital lobe, left posterior cingulate gyrus, left and right precunei) where AD-related hypoperfusion was found in the previous step, and two regions (right occipital lobe and right parietal lobe) where hypoperfusion was expected based on previous reports.

There were statistical significant differences between AD cases and controls in left occipital lobes, left posterior cingulate gyri, left and right precunei (p < 0.05). The area under the receiver operating characteristic curve for the classification of AD was 0.903 based on a leave-one-out-by-case test.

Yamashita, Y, Arimura, H, Yoshiura, T, Tokunaga, C, Monji, A, Toyofuku, F, Kuwazuru, J, Magome, T, Nakamura, Y, Ohya, N, Honda, H, Ohki, M, Computer-aided Classification System for Alzheimer's Disease Based on Functional Image Features of Arterial Spin Labeling Cerebral Blood Flow.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11009146.html