A direct sinogram-restoration method for fast image reconstruction in compact DOI-PET systems

In a recent study, the authors proposed a new PET design concept that integrates hardware design and data processing for improvement of PET technology. In particular, they the use of new-generation depth-of-interaction in hardware for generation of redundant, yet imperfect, information, together with sinogram restoration (SR) techniques in software that makes efficient use of the redundant information available for correction of data imperfection. The authors demonstrated that this design approach can be used to build PET systems of compact geometry that has the potential of offering high sensitivity, high resolution, and low cost. In this work, the authors continue their development by introducing a direct SR method specifically targeted for design of dedicated small-animal systems. Compared to iterative algorithms considered in the authors' previous study, this direct approach requires essentially no computation time, can be implemented in hardware/firmware, and may be suitable for on-the-fly processing of PET data. Performance of the direct method was evaluated by computer simulation studies. The authors' results indicate that, when scatter and randoms are excluded from simulated data, the proposed method can effectively correct for the parallax errors that exist in the raw data of a compact PET systems, thereby producing images of quality comparable to those generated by a conventional larger-ring system that employs conventional data processing. This observation, however, may be affected by the presence of scatter and randoms, an important issue to be addressed in the authors' future work.