The broad, long-term objective of the proposed research is to develop Bayesian statistical methods for emission tomographic reconstruction that will lead to improvements in quantitation and lesion detection. Our application is for SPECT regional cerebral blood flow (rCBF) imaging, and our methods have potential impact in the detection and localization of epileptic seizure foci, as well as application to stroke and dementia. One theme is the incorporation of new types of a priori information into the reconstruction by means of Bayesian methods. A second theme concerns new algorithms that can be used to compute improved reconstructions given the optimization problems that result from these Bayesian formulations. Two types of prior information will be modelled: (l) the local spatial character of the underlying rCBF pattern, including the observation that the pattern is comprised of approximately piecewise linear (ramplike) regions and piecewise constant (flat) regions; and (2) the correlation of anatomical boundaries, as seen in coregistered MR scans, with rCBF boundaries. A feature that distinguishes this work is the use of primate (rhesus) rCBF autoradiographic data to derive anatomically and functionally correct phantoms. A second factor is our use of new mathematical """"""""weak plate"""""""" priors that are able to model piecewise linear regions. A third novel feature is our introduction of a new class of reconstruction algorithms, based on continuation methods, that yield improved image quality relative to images computed from the same projection data, but by conventional means. The new priors and algorithms will be able to lower noise, decrease resolution distance, and improve detection of low contrast regions. These improvements can support a variety of clinical tasks. In addition, the lower ensemble variance of our reconstructions along with low bias will yield improved accuracy and reproducibility of quantitative information. We shall evaluate and compare our new methods with other competing methods according to several task performance criteria. In addition, we shall evaluate the utility of our methods in a task of localizing epileptic seizure foci from SPECT ictal and interictal clinical data. In this clinical application, our new reconstruction methods shall be applied retrospectively to a database of projection data acquired from SPECT ictal and interictal scans. Our methodology is applied specifically to rCBF SPECT, but is readily generalizable to PET metabolic brain imaging and SPECT neuroreceptor brain imaging, thus extending its potential clinical impact.

Agency
National Institute of Health (NIH)
Institute
National Institute of Neurological Disorders and Stroke (NINDS)
Type
Research Project (R01)
Project #
1R01NS032879-01A1
Application #
2271368
Study Section
Special Emphasis Panel (ZRG7-SSS-X (41))
Project Start
1995-06-01
Project End
1999-05-31
Budget Start
1995-06-01
Budget End
1996-05-31
Support Year
1
Fiscal Year
1995
Total Cost
Indirect Cost
Name
State University New York Stony Brook
Department
Radiation-Diagnostic/Oncology
Type
Schools of Medicine
DUNS #
804878247
City
Stony Brook
State
NY
Country
United States
Zip Code
11794
Hsiao, Ing-Tsung; Khurd, Parmeshwar; Rangarajan, Anand et al. (2006) An overview of fast convergent ordered-subsets reconstruction methods for emission tomography based on the incremental EM algorithm. Nucl Instrum Methods Phys Res A 569:429-433
Hsiao, Ing-Tsung; Rangarajan, Anand; Gindi, Gene (2003) A new convex edge-preserving median prior with applications to tomography. IEEE Trans Med Imaging 22:580-5
Wang, W; Gindi, G (1997) Noise analysis of MAP-EM algorithms for emission tomography. Phys Med Biol 42:2215-32