Efficient Monte Carlo based reconstruction for general quantitative SPECT

Accuracy of SPECT images improves if photon scatter is modeled more accurately during reconstruction. Monte Carlo simulation (MCS) is a general method for detailed modeling of scatter, but to date, fully 3D MCS-based statistical reconstruction approaches have been prohibitively slow and require internal computer memory of hundreds of Giga byte. Here, we develop computationally efficient fully 3D MCS-based reconstruction strategies by combining the following methods: (i) a dual matrix reconstruction approach is used to accelerate the reconstruction and avoid massive transition matrix pre-calculation and storage, (ii) stochastic photon transport calculation in MCS is combined with an analytic detector modeling step, in order to achieve low noise in the Monte Carlo based re-projection already after only a small number of photon histories have been tracked, and (iii) the number of photon histories simulated is kept an order of magnitude lower in early iterations than in the last couple of iterations, or alternatively, the major part of the photon histories calculated for an early iteration is re-used in later iterations. The typical reconstruction time is about half an hour on a dual processor PC. Since MCS can calculate photon transport for any clinically used photon energy or patient attenuation distribution, the proposed methodology is expected to be an important tool for facilitating general absolute quantitative SPECT reconstruction within clinically acceptable computation times