Correntropy coefficient analysis of fMRI using reference model

A non-invasive functional MRI (fMRI) has emerged effective in the investigation of the functionality of human brain. However, detection of functional activation is often complicated by the presence of noise. In this paper, we propose an approach based on correntropy, a recently introduced measure which incorporates both amplitude and temporal structure characteristics of time series in single functional measure. Using correntropy coefficient as the test statistic, nonparametric approach is carried out via pre-whitening resampling transform to calculate the statistical p-values. Experimental results suggest that proposed method enables more effective brain activation detection compared with mutual information.