Meta-analysis of functional neuroimaging data

Abstract

Functional neuroimaging offers huge amounts of data that require computational tools to help extract useful information about brain function. The ever increasing number of neuroimaging studies (above 5000 in 2012 alone) suggests the need for a meta-analysis of these findings. Meta-analysis is aimed at increasing the power and reliability of findings from individual studies. Currently, two methods of meta-analyses are the most popular in brain imaging literature. The coordinate based meta-analysis (CBMA) which refers to the maximum likelihood of brain activation based on a universal three dimensional coordinate system. The image based meta-analysis (IBMA) which considers the effect sizes from different studies to increase statistical power ignoring the inter-study consistency requirements. This technique is, however, suitable to account for inter-subject variability either pooled over studies or including the inter-study variability. While the coordinate based meta-analysis is easily found through published literature, the image based analysis requires the statistical parametric maps available. These Data mining techniques applied in brain imaging is often termed as the new paradigm in cognitive neuroscience. We here discuss in detail about the available analysis methods.