Sparse recovery in myocardial blood flow quantification via PET

In this paper we are considering the problem of myocardial blood flow quantification via dynamic positron emission tomography (PET). In dynamic PET the measured data is divided into small temporal bins leading to a low signal-to-noise ratio (SNR) in each temporal bin. Thus, the physiological parameters have to be estimated using bad quality reconstructions. We want to overcome this problem by incorporating apriori information in the form of a linear physiological model, represented by basis functions. To identify the parameters in question we combine the reconstruction process with sparsity regularization.