Dose optimization of SPECT-MPI reconstruction algorithms for perfusion-defect detection

Abstract

In SPECT-myocardial perfusion imaging (MPI), as in any imaging modality that involves ionizing radiation, minimization of radiation dose to the patient is an important objective. This paper reports the first of a series of studies we are performing to optimize the dose for various reconstruction algorithms, the ultimate aim being to provide dose recommendations, corresponding to optimized algorithm parameters, and performance comparisons across algorithms. Dose optimization involves determination of the extent to which administered activity can be reduced without sacrificing diagnostic accuracy. An important consideration is to conduct the optimizations in a principled way so as to avoid overfitting of the algorithm parameters during optimization. Using clinical SPECT-MPI data with realistic simulated lesions we optimized each reconstruction algorithm for maximum perfusion-defect detection performance, in which each study was evaluated for the presence of perfusion defects by using the total perfusion deficit score computed by the clinically validated Quantitative Perfusion SPECT (QPS) software package. Preliminary results suggest that tracer dose (and hence radiation dose) can be greatly decreased from current clinical dose levels for ordered subsets expectation-maximization OS-EM without significant loss of diagnostic performance compared to reconstruction with filtered-back projection (FBP) and imaging at current dose levels.