Quantitative evaluation of information loss for Compton cameras

Compton cameras decouple the inverse relationship between spatial resolution and detection sensitivity which compromises the performance of conventional collimated cameras. However, this improvement is usually achieved at the expense of the amount of information conveyed by each detected photon. In this paper, we propose a simple approach to calculate the information loss for a ring Compton camera. We describe this information loss as "decoding penalty", defined as the ratio of the variance of reconstructed intensity for a pixel of interest for a ring Compton camera to that for a mechanically collimated camera normalized on a per-detected-photon basis. The uniform Crame/spl acute/r-Rao bound, our mathematical tool, provides a lower bound on the variance that is dependent on the system and data statistics alone, rather than the estimator. The results suggest that ring Compton cameras perform comparably to conventional collimated cameras at an incident photon energy of 140 keV and substantially outperform their counterparts at 364 keV.