IEEE NUCLEAR SCIENCE SYMPOSIUM - CONFERENCE RECORD
Nuclear Science Symposium/Medical Imaging Conference location:Fajardo, PR date:23-29 October 2005
High resolution functional imaging of small animals is often obtained by single pinhole collimated SPECT with a circular acquisition orbit. However, the acquired data are incomplete. Multipinhole SPECT has the advantage of adding more information due to its improved sampling and possibly increases the resolution and the sensitivity. To evaluate different pinhole collimator designs a method is needed that tells how much information is missing, or equivalently, how well the animal can be reconstructed. In this paper we propose two methods to facilitate the evaluation process for single and multipinhole collimator designs. Firstly, we examine the quality of each post-processed MLEM reconstructed voxel by studying its linearized local impulse response (LLIR) and its (co)variance for a predefined target resolution. Secondly, since increasing the number of pinhole apertures may cause the projections to overlap, point and circular artifacts can occur in the reconstruction image. The position of these 'ghost points' and 'ghost circles' can be well predicted analytically. We also give some rules of thumb to estimate the (relative) intensity of the artifacts. The methods are validated with simulations.