Monte-Carlo-Based Scatter Correction for Quantitative SPECT Reconstruction

Quantitative SPECT as well as simultaneous acquisition of multiple isotopes with SPECT in the clinical field, although clinically interesting, are still limited by reconstruction artifacts and computing power. As a considerable step in this direction, we have implemented an efficient reconstructor with variance reduced Monte-Carlo-simulation in the forward and/or backward projection of an OS-EM iteration. Apart from a quantitatively accurate scatter estimation the integrated MC simulation allows us to include all effects that are relevant for the multiple isotope task. The reconstruction problem is explicitly formulated as a combined maximum-likelihood estimation for all unknowns in one step and implemented efficiently on the basis of the OS-EM algorithm using the Effective Scatter Source approach. The algorithm has been evaluated on simulated as well as measured phantom data. Two isotopes (Tc99m, Tl201) were tested on the voxelized NCAT phantom and evaluated quantitatively. In addition we performed phantom measurements to evaluate the method on a Philips CardioMD with a Vantage line source for attenuation correction. A clear advantage of the proposed approach is its robustness and generalizability. It is currently being evaluated in clinical applications like simultaneous dual isotope cardiac imaging.