To date, all single photo imaging systems in Nuclear Medicine use mechanical collimation to form an image of the radioisotope distribution on the face of a gamma camera. These collimators impose a hard physical constraint that couples detection sensitivity and spatial resolution in an inverse relationship: for a fixed field of view, one can improve the resolution by a factor of two only by reducing sensitivity a factor of four. Furthermore, collimator performance in general degrades substantially as gamma-ray energy increases. The focus of this proposal is to determine the practicality of Compton camera imaging for SPECT and experimentally characterize its performance. In contrast to mechanical collimation, both spatial resolution and noise equivalent sensitivity can be increased. Demonstration of practicality requires approaches be developed to image reconstruction for an imaging system with a system matrix that can easily approach 10E+10 bytes. Methods will be investigated to reduce this matrix size that permit approximate analytical solutions to image reconstruction and will also employ list- mode likelihood as an exact, statistically based algorithm. Noise equivalent performance for various system configurations will be determined using uniform bound techniques which characterize performance limits based only on the system matrix and a given object. Monte Carlo studies will be used to develop and test analytical system models and also to generate simulated data sets for testing image reconstruction and confirming inferences drawn from bound calculations. The Compton camera module will be constructed using multiple silicon pad detectors read out with self-triggered multi-channel integrated circuits and intelligent read-out circuitry. It will be assembled with a ring scintillation detector to form a Ring Compton Camera (RCC) to demonstrate that a system based on the Compton imaging principles is markedly superior for SPECT imaging.
