Functional magnetic resonance imaging (fMRI) data is used to analyze the interactions between brain regions. A full image of the brain (consisting of about 10^6 voxels) is recorded every few seconds resulting in a high-dimensional time series. Each observation of the time series consists of a full picture, which we will model as a function. The examination of the covariance structures between brain regions without assuming the specifications of the experiment to be known is of particular interest. Therefore, we propose asymptotic nonparametric testing procedures in order to detect changes in the covariance structure of functional time series. We apply resampling procedures to get a more stable change point procedure.