Three-dimensional image reconstruction methods for quantitative myocardial perfusion SPECT

We have developed three-dimensional (3D) image reconstruction methods for quantitative myocardial SPECT based on accurate model of the imaging process and statistical iterative reconstruction algorithms. The methods incorporate information about the non-uniform attenuation in the patient's chest region, and 3D models of the spatially variant collimator-detector and scatter responses in the iterative algorithm. Increasingly accurate 3D reconstructed images are obtained as the degrading factors are progressively compensated through increasing number of iterations. The final image estimate is obtained when an optimum trade-off between accuracy and noise is achieved. Recent advances include accurate determination of the attenuation distribution in the chest region using transmission CT methods and optimization of the implementation of the complex collimator-detector and scatter response functions in the iteration process. Taking advantage of the recently available fast iterative OS-EM algorithm and computer hardware, the 3D image reconstruction methods can be implemented in clinically acceptable processing time.