The objective of the proposed research is to develop and apply accurate scatter models to facilitate the utilization of energy information in single photon emission computed tomography (SPECT) for improved quantitation and image quality. These models are based on the physics of the scattering process. In current practice, photons with energies below the photopeak window are either unused or used only to estimate and subtract the scatter component in the photopeak window. The reason for this is that raw images from lower energy windows contain a large scatter component and have increasingly poor contrast as compared to that from the photopeak energy window. In this study we propose to develop accurate methods for modeling the scatter component in these lower energy windows and to apply them in three different ways to improve the diagnostic utility of SPECT. Specifically, we propose to (I) extend and develop accurate scatter models to allow reconstruction of activity distributions from projection data acquired in multiple energy windows; (2) develop a method, based on a theoretical model relating the single-scatter distribution to the (nonuniform) attenuation distribution, for simultaneously reconstructing the attenuation and attenuation compensated activity distributions; and (3) develop methods for modeling and compensating for cross-talk in images reconstructed from simultaneously acquired dual radionuclide projection data.
In specific aim (1) we propose to use, in addition to the scattered and primary photons in the photopeak energy window, the (usually discarded, predominantly scattered) photons in lower energy windows to reduce statistical image noise. As a result of using the accurate scatter model in the reconstruction process, this reduction in noise would be accomplished with little loss of image contrast. In other words, this method would achieve the contrast improvement obtained using conventional scatter compensation methods, but with lower image noise than uncompensated images.
In specific aim (2) we propose to use data from multiple energy windows to estimate the single-scatter distribution in the projection data. The single-scatter distribution, in combination with an analytical scatter model, will then be used to estimate the electron density, and from it the attenuation distribution.
In aim (3) we propose to study and develop methods to compensate for the cross-talk between simultaneously acquired dual radionuclide studies. The simultaneous acquisition of these studies would reduce the acquisition time required to perform separate studies. Examples of clinical applications are rest/stress cardiac perfusion SPECT and brain SPECT studies. We propose to accomplish this by studying the cross-contamination using a combination of experimental measurements and Monte Carlo simulations. We will develop methods based on these studies to reduce cross-contamination between projection data for the different radionuclides. The methods developed to accomplish the three specific aims will be evaluated using MC simulation, experimental phantom studies and patient data using both quantitative measures of image quality and observer experiments.
| Kadrmas, D J; Frey, E C; Tsui, B M (1999) Simultaneous technetium-99m/thallium-201 SPECT imaging with model-based compensation for cross-contaminating effects. Phys Med Biol 44:1843-60 |
| Kadrmas, D J; Frey, E C; Karimi, S S et al. (1998) Fast implementations of reconstruction-based scatter compensation in fully 3D SPECT image reconstruction. Phys Med Biol 43:857-73 |
| Kadrmas, D J; Frey, E C; Tsui, B M (1998) Application of reconstruction-based scatter compensation to thallium-201 SPECT: implementations for reduced reconstructed image noise. IEEE Trans Med Imaging 17:325-33 |
| Kadrmas, D J; Frey, E C; Tsui, B M (1997) Analysis of the reconstructibility and noise properties of scattered photons in 99mTc SPECT. Phys Med Biol 42:2493-516 |
