In this renewal application, we continue our investigation of the fundamental limits on quantitative SPECT imaging and expand our scope to include PET. We complete our efforts in the area of special purpose collimation for brain SPECT imaging, and add new emphases in the areas of cardiac SPECT imaging and PET time-of-flight (TOF) imaging. Our goals remain two-fold: to develop quantitative task based metrics for system assessment, and to use these metrics to quantify the value of improvements to ECT imaging systems. In the proposed project period, we will focus on the most fundamental aspect of the imaging system, data acquisition, which determines the ultimate limits on quantitation. We will assess the merits of three recent hardware advances in terms of performance in estimation tasks related to Alzheimer disease, Parkinson disease, and cardiovascular disease. The brain and cardiac hardware improvements are each expected to provide about an order of magnitude improvement in sensitivity without compromising spatial resolution. Similarly, the addition of TOF information to PET imaging promises to provide substantial gains in image quality. It is extremely uncommon in relatively mature fields of medical imaging to discover new hardware modifications which promise to provide such a significant improvement. In the proposed research we will quantify the gains in estimation task performance to be expected for quantitative brain and cardiac imaging from these three advances, one of which emerged from the previous project period. We will manufacture a collimator, designed during the previous project period, which is expected to increase sensitivity of brain SPECT by, on average, a factor of 10. We will also assess the performance of a high-sensitivity dedicated cardiac SPECT system, and quantify the advantages of PET TOF information for both brain and cardiac applications. Bias and precision of estimates of activity concentration and kinetic parameters will be determined by analysis, simulation, and experiment. A range of clinically realistic noise levels will be considered, including noise levels at which we have previously shown that efficient estimation of nonlinear parameters is not possible.
Showing the most recent 10 out of 20 publications
