Information optimization of projective tomographic imaging systems

A mutual information (MI) criterion for the evaluation of tomographic imaging systems is introduced. The MI quantifies the quality of the tomographic projections in terms of the transfer of information from a slice of a three-dimensional object to the raw data at the detectors, and it provides a relevant tradeoff between sensitivity and resolution. The authors consider the application of the MI criterion to aperture evaluation for one-dimensional linear and three-dimensional parallel slice single photon emission computed tomography (SPECT) geometries. For the ideal case of known emission times, analytic expressions that give a necessary and sufficient condition for an aperture to maximize MI can be derived. Otherwise, an upper bound on the MI can be derived. The authors present a numerical comparison between the MI of several popular aperture designs and the MI-optimal apertures presented here