The broad goal of this research is to improve the diagnostic accuracy of clinical positron-emission tomography (PET) scans and the quantitative accuracy of research scans by accurately correcting for the effects of Compton scattering. Current methods of correcting for Compton scatter are lacking in three respects: (1) They assume that the scattering medium is homogeneous; (2) They treat scattering as a two-dimensional phenomenon, ignoring scatter from adjacent slices; and (3) They are empirically based, despite the fact that the physics of Compton scatter are well known. These shortcomings are more pronounced in cardiac scans, the scans that are currently of greatest clinical interest, for two reasons: First, scatter is a major source of image degradation in the chest cause of the relatively long path-lengths encountered there; and second, the lungs introduce significant heterogeneities in the scattering medium. The proposed method, termed model-based scatter correction, makes no assumptions about the scattering medium, takes the three-dimensional nature of scatter into account and is grounded in a model of the fundamental physics of Compton scatter. By appropriately formulating the method as a ray-tracing algorithm, the computational complexity is reduced by a factor of n over what might be expected, thereby making the algorithm computationally practical on currently available processors. A preliminary version of this algorithm has been developed based on several restrictive assumptions. Initial phantom studies suggest that a fully developed version of the algorithm will perform significantly better scatter correction than current approaches.
The specific aims of this proposal are to: (1) Develop the preliminary version to the point that it can be used clinically; (2) Evaluate the performance of this method; (3) Extend the algorithm for use with fully 3D PET systems; (4) Evaluate the performance of the 3D model; and (5), Formulate the algorithm for use on high-speed processors. Methods include the use of mathematical modeling, phantom studies, and previously acquired clinical PET data. This project will be performed in cooperation with the departments of Neurology, Radiation Sciences, and Internal Medicine.

Agency
National Institute of Health (NIH)
Institute
National Cancer Institute (NCI)
Type
Research Project (R01)
Project #
5R01CA054959-03
Application #
2096231
Study Section
Diagnostic Radiology Study Section (RNM)
Project Start
1991-07-01
Project End
1995-03-31
Budget Start
1993-07-01
Budget End
1995-03-31
Support Year
3
Fiscal Year
1993
Total Cost
Indirect Cost
Name
Washington University
Department
Biostatistics & Other Math Sci
Type
Schools of Medicine
DUNS #
062761671
City
Saint Louis
State
MO
Country
United States
Zip Code
63130
Ollinger, J M (1996) Model-based scatter correction for fully 3D PET. Phys Med Biol 41:153-76