fMRI binary detection of brain activated regions with graph-cuts

The functional Magnetic Resonance Imaging (fMRI) is a technique with increasing applications in studying the brain function. The blood-oxygenation-level-dependent (BOLD) is a fMRI method that allows the detection of brain activated regions after the application of an external stimulus, e.g., visual or auditive. This technique is based on the assumption that the metabolism increases in activated areas as well as the oxygen uptake. Analising this information is a challenging problem because the BOLD signal is very noisy and its changes due to the application of a stimulus are very weak. Therefore, the detection of temporal correlations with the applied stimulus requires sophisticated statistical algorithms to understand if the changes on the BOLD signal are pure noise or are related with the applied stimulus, called paradigm in the fMRI scope.