ABSTRACT
We present the fusion of anatomical data as a method for improving the reconstruction in single photon emission computed tomography (SPECT). Anatomical data is used to deduce a parameterized model of organs in a reconstructed slice using spline curves. This model allows us to define the imaging process, i.e., the direct problem, more adequately, and furthermore to restrict the reconstruction to the emitting zones. Instead of the usual square pixels, we use a new kind of discretization pixel, which fits to the contour in the region of interest. In the reconstruction phase, we estimate the activity in the emitting zones and also the optimum parameters of our model. Concentrating on the left ventricular (LV) wall activity, the simulation and phantom results show an accurate estimation of both the myocardial shape and the radioactive emission.
