The polyvertex score (PVS): a whole-brain phenotypic prediction framework for neuroimaging studies

Abstract

The traditional brain mapping approach has greatly advanced our understanding of the localized effect of the brain on behavior. However, the statistically significant brain regions identified by standard mass univariate models only explain minimal variance in behavior despite increased sample sizes and statistical power. This is potentially due to the generalizable explanatory signal in the brain being non-sparse, therefore not captured by the thresholded, localized model. Here we introduced the Bayesian polyvertex score (PVS-B), a whole-brain prediction framework that aggregates the effect sizes across all vertices to predict individual variability in behavior. The PVS-B estimates the posterior mean effect size at each vertex with mass univariate summary statistics and the correlation structure of the imaging phenotype, and weights the imaging phenotype of participants from an independent sample with these posterior mean effect sizes to estimate the generalizable effect of a brain-behavior association. Empirical data showed that the PVS-B was able to double the variance explained in general cognitive ability by an n-back fMRI contrast when compared to prediction models based on the mass univariate parameter estimates as well as models in which only vertices thresholded based on p-value were included. A fivefold improvement in variance explained by the PVS-B was observed using a stop signal task fMRI contrast to predict individual variability in the stop signal reaction time. We believe that the PVS-B can shed light on the multivariate investigation of brain-behavioral associations and will empower small scale neuroimaging studies with more reliable and accurate effect size estimates.