Accurate image reconstruction using DOI information and its implications for the development of compact PET systems

We study the use of a computationally efficient sinogram restoration (SR) approach for reconstruction of 2D PET data acquired by systems with depth-of-interaction capability. When the effects of scatter and randoms are ignored, our preliminary results show that the proposed approach can yield images of excellent resolution uniformity for a hypothetical compact system with a field-of-view (FOV) as large as 98% of the detector-ring size. In comparison with images generated by the conventional rebinning approach, the resolution uniformity across the FOV in the image produced by incorporating SR is improved by a factor of about 10, and the spatial resolution at the edge of the FOV is improved by a factor of about 2.2 Moreover, the SR approach produces better intensity recovery and source positioning. Integration of the use of DOI detectors and SR reconstruction may thus provide a new way for developing compact PET systems. For a given FOV, such a compact system can produce images of quality better than those obtained by a conventional system that employs a detector ring approximately 50% larger than the compact system. The much reduced detector-ring size of the compact system can greatly reduce production cost, improve image resolution and resolution uniformity, and increase system sensitivity.