Dunedin Brain Imaging Study MRI Protocol
Brain Age
We generated brain-age scores using a recently published, publicly available algorithm (Liem et al. 2017). Unlike some of the other publicly-available brain-age algorithms, this method uses a stacked prediction algorithm based on multiple measures of brain structure including cortical thickness, surface area, subcortical and global brain volume all derived from Freesurfer version 5.3 (Fischl 2012). Test-retest reliability was assessed in 20 Dunedin Study members (mean interval between scans = 79 days). The ICC of brain-age was .81, indicating excellent reliability (Cicchetti 1994). Moreover, we chose this algorithm because of its performance in predicting chronological age in independent samples and its sensitivity to age-related cognitive impairment in old age (Liem et al. 2017). Data from six study members were excluded due to major incidental findings or previous head injuries (e.g., large tumors or extensive damage to the brain). This resulted in brainAGE data from 869 Study members.
References
Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment. 1994; 6: 284–290.
Fischl B. FreeSurfer. Neuroimage 2012; 62: 774–781.
Liem F, Varoquaux G, Kynast J, Beyer F, Kharabian Masouleh S, Huntenburg JM et al. Predicting brain-age from multimodal imaging data captures cognitive impairment. Neuroimage 2017; 148: 179–188.