Understanding brain structure and function can benefit from studying functional connectivity. A common methodology to measure functional connectivity between two brain regions is to estimate the correlation between their corresponding average time courses. Usually, these correlations are computed either via the Pearson estimator or the non-parametric Spearman estimator. However, these two measures do not fully reflect the information we want to extract about the spontaneous activity in the different areas of the brain. In this paper, we propose to estimate functional connectivity between two regions by modeling the activation parts of the time course as the extreme events and by measuring the co-activation between these events. We show that our new measure of functional connectivity contains key information about the co-activations, which is lost when using common functional connectivity measures; i.e., Pearson or Spearman correlation.