Dynamic objects in emission tomography reconstructed with temporal B-splines

In medical imaging the role of 4D dynamic reconstruction methods are fundamental, for example for the study of moving objects, like heart or lung motion, or tracer activity (kinetics). Many existing reconstruction methods in nuclear medicine, e.g. PET, are based on Poisson modelling, however often the objects are static and motion is not taken into account. We make use of a model for inhomogeneous poisson point processes and will study the relation between different methods, that reconstruct data from a given source, in the context of motion. The unknown point sources will be approximated with temporal b-splines and the effect of regularisation will be investigated for different orders of b-splines.