Improved non-invasive quantification of physiological processes with dynamic PET using blind system identification

The traditional quantitative study using tracer kinetic modeling requires repeated blood sampling to measure the tracer concentration in plasma. It is invasive, time-consuming and costly. We propose a new approach to estimate physiological parameters for dynamic Positron Emission Tomography (PET) without taking blood samples. With the new approach, the quantification problem is first converted to a discrete blind system identification problem. A multi-channel blind identification technique is applied to estimate the required system parameters. A Monte Carlo simulation is carried out for the proposed approach on estimating the regional cerebral metabolic rate of glucose (rCMRGlc) with the fluoro-deoxy-2-glucose (FDG) model. The results show that the proposed approach can estimate the required physiological parameters in a comparable manner to that of traditional invasive approaches.