• Hidemasa Takao, Naoto Hayashi, and Kuni Ohtomo.
• Department of Radiology, Graduate School of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, Japan. Electronic address: takaoh-tky@umin.ac.jp.
• Magn Reson Imaging. 2015 Jul 1; 33 (6): 816-21.

AbstractThe identification of individual differences in brain morphology is important to understand the background of individual differences in brain functions. In the present study, we investigated whether brain morphology is discernibly different among individuals and is personally identifiable information. Using structural magnetic resonance imaging data from 215 healthy subjects scanned twice (scan interval = 1.0 ± 0.1 years), we performed brain recognition by image normalization using a voxel-based morphometry approach, feature extraction based on principal component analysis, and calculating the Euclidean distances between image pairs projected into the subspace. Even with only 32 dimensions used for projection, the rank-one identification rate was 99.5%. With ≥112 dimensions used, the rank-one identification rate was 100%. At a false accept rate of 0.01%, the genuine accept rates were 95.8% and 100% with 32 and ≥128 dimensions used for projection, respectively. There was little difference in the Euclidean distances among different combinations of scanners used or between probe-gallery image pairs with and without scanner upgrade. These results indicate that brain morphology can identify a specific individual; i.e., brain morphology is personally identifiable information. Individually different brain morphology may occur as a collection of differences in brain structures that reflect individual differences in a variety of performances and various psychological characteristics and behavior patterns, and may provide the background of individual differences in personality and brain function.Copyright © 2015 Elsevier Inc. All rights reserved.

23 keywords...

hide…