A super-resolution feasibility study in small-animal SPECT imaging

Lack of spatial resolution is a major drawback in small-animal SPECT imaging, particularly when parallel hole collimators are employed. This work evaluates the feasibility of enhancing the spatial resolution of a small-animal SPECT scanner by applying an approach based on a super-resolution technique combined with dedicated hardware. The detectors are based on PS-PMTs (Hamamatsu, H8500), NaI(Tl) crystal arrays and parallel hole collimators. The system enables the mechanical shifting of the gamma-cameras and the object under study during the acquisition in steps smaller than the detector intrinsic sampling pitch. This fact allows using a super-resolution technique without the necessity of blind movement estimators between projections. In the absence of noise the algorithm converged to the original image, and with real noisy data it achieved improvements in resolution between 28% (tomographic) and 42% (planar) in phantom studies.