Reproducibility of automated brain morphometric estimates derived from multi-site structural MRI studies
Busa E, van der Kouwe A, Fischl B, Dale A, Jovicich J, Han X, Czanner S, Greve D, Pacheco J
1MGH/MIT/HMS Athinoula Martinos Center for Biomedical Imaging
Neuroimaging longitudinal and multi-site studies of change over time are becoming increasingly a standard element of clinical neuropsychiatric research. These studies create the imperative to characterize and correct technological sources of variance that limit image reproducibility in high-resolution structural MRI studies. Here we extend previous work to study the variability of morphometry results, cortical thickness and subcortical volume estimates, using data from a group of 5 healthy volunteers that were scanned twice at each of five different sites having clinical 1.5T systems (GE, Siemens, Picker). The measured cross-site variability was under 10%. Gradient non-linearity correction did not significantly reduce it.