Advocates of using imaging for genetic analysis pointed to several advantages that imaging provides now, as well as several potential advantages that it could provide in the foreseeable future. First, neuroimaging can yield reproducible measures of brain structure and (perhaps to a lesser extent) brain function. Structural measures of the brain, from MRI, tend to have relatively high reproducibility across measurement methods, and are generally consistent with expert tracings of the same structures (see Supplement of Stein et al. 2012; many studies have investigated the reliability of measures from brain MRI, e.g., Pengas et al. 2009). In a recent GWAS analysis, Holmes et al. (2012) showed high reliability for automated brain volumes of hippocampus (r = .98), amygdala (r = .91), and intracranial volume (r = .99) for a cohort of data collected across investigators and matched scanners. Nugent et al. (2012) also studied the inter-scanner reliability of the FIRST software for segmentation (Patenaude et al. 2011), and found it to be high. However, it is overly optimistic to always expect high reproducibility from automated segmentations of brain MRI, and
