Functional Magnetic Resonance Imaging (fMRI) of the brain gives information about brain activity over time. In experiments, a subject performs a particular task and neuroscientists wish to identify "components" of brain activity associated to that task.

I will briefly introduce the problem we wish to solve, in both biological and mathematical terms, and will describe an analysis method which we are currently investigating. This method assumes some sparsity of the components of activity and makes use of an iterative algorithm of Daubechies, Defrise & De Mol to solve the variational problem we pose.