DESCRIPTION (provided by applicant): The objective of the proposed research is to develop a rigorous theoretical and experimental framework for the objective assessment of image quality when the tasks in question are medically relevant estimation tasks. We propose to apply this framework to compare different configurations of a specific SPECT (single photon emission computed tomography) imaging system being developed by our collaborators and to optimize the parameters of this system. We will focus on the task of locating small tumors or other focal objects such as sentinel lymph nodes and the task of estimating integrated tumor activity. We envision three stages in the development of a system optimized for a specific task such as tumor-location estimation. The first stage is the simulation of the entire system to determine optimal parameters. The second stage is the validation of the design by building a small system and using it for phantom or small-animal imaging studies. The final stage is to build a full-sized system for human studies. The work we propose will develop methods for addressing the first stage in this process and will result in the ability to design medical imaging systems that maximize observer performance on estimation tasks. For the second stage, we will apply these methods to the tasks of estimating tumor location and integrated tumor activity using newly designed phantoms and experimental data from a small-animal SPECT system. We will not address the third stage of system optimization in this proposal. We will treat estimation tasks in a general, mathematical sense that will allow us to rapidly extend our work to other imaging modalities (including human imaging systems) and to other estimation tasks without difficulty. Ultimately, this research will result in imaging systems that improve the diagnosis and treatment of patients.
|Effective start/end date||8/1/04 → 7/31/08|
- Biochemistry, Genetics and Molecular Biology(all)
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.