An unbiased data-driven age-related structural brain parcellation for the identification of intrinsic brain volume changes over the adult lifespan

Publication date: 1 April 2018
Source:NeuroImage, Volume 169
Author(s): Epifanio Bagarinao, Hirohisa Watanabe, Satoshi Maesawa, Daisuke Mori, Kazuhiro Hara, Kazuya Kawabata, Noritaka Yoneyama, Reiko Ohdake, Kazunori Imai, Michihito Masuda, Takamasa Yokoi, Aya Ogura, Toshihiko Wakabayashi, Masafumi Kuzuya, Norio Ozaki, Minoru Hoshiyama, Haruo Isoda, Shinji Naganawa, Gen Sobue
This study aims to elucidate age-related intrinsic brain volume changes over the adult lifespan using an unbiased data-driven structural brain parcellation. Anatomical brain images from a cohort of 293 healthy volunteers ranging in age from 21 to 86 years were analyzed using independent component analysis (ICA). ICA-based parcellation identified 192 component images, of which 174 (90.6%) showed a significant negative correlation with age and with some components being more vulnerable to aging effects than others. Seven components demonstrated a convex slope with aging; 3 components had an inverted U-shaped trajectory, and 4 had a U-shaped trajectory. Linear combination of 86 components provided reliable prediction of chronological age with a mean absolute prediction error of approximately 7.2 years. Structural co-variation analysis showed strong interhemispheric, short-distance positive correlations and long-distance, inter-lobar negative correlations. Estimated network measures either exhibited a U- or an inverted U-shaped relationship with age, with the vertex occurring at approximately 45–50 years. Overall, these findings could contribute to our knowledge about healthy brain aging and could help provide a framework to distinguish the normal aging processes from that associated with age-related neurodegenerative diseases.



http://ift.tt/2yMKUTV