A multigrid maximum-likelihood image-reconstruction algorithm for positron emission tomography

Miscellaneous · Sun Jan 01 00:00:00 EST 1989

OSTI ID:7036720

Ranganath, M V

The problem of image reconstruction in positron emission tomography (PET) requires estimating the number of photon pairs emitted from the radioactive source deposited within the body. Using the maximum likelihood principle, this problem may be reduced to determining an estimate of emitter density that maximizes the probability of observing the actual detector count data over all possible emitter density distributions. All the reconstruction algorithms that have been proposed to date using this method attempt to solve the problem at a fixed discretization level or fixed grid size. However, restraint to a fixed grid size results in slow convergence and computational inefficiency. It is hypothesized that the use of a hierarchy of several grids at different discretization levels to solve the image reconstruction problem in PET will improve the performance of the reconstruction algorithm and provide a better resolution image for the same computational effort. To verify this hypothesis, a new reconstruction algorithm called the multi-grid expectation maximization (MGEM) algorithm has been developed. The performance of this algorithm is compared with the algorithm based on a fixed grid size, termed the single-grid expectation maximization (SGEM) algorithm. Studies done on simulated phantom data and real PET camera data support the hypothesis.

• Research Organization: Houston Univ., TX (United States)
• OSTI ID: 7036720
• Resource Relation: Other Information: Thesis (Ph.D.)
• Country of Publication: United States
• Language: English

Related Subjects

62 RADIOLOGY AND NUCLEAR MEDICINE
POSITRON COMPUTED TOMOGRAPHY
ALGORITHMS
COMPUTER CALCULATIONS
DATA ANALYSIS
EXPERIMENTAL DATA
POSITRON CAMERAS
STATISTICAL MODELS
CAMERAS
COMPUTERIZED TOMOGRAPHY
DATA
DIAGNOSTIC TECHNIQUES
EMISSION COMPUTED TOMOGRAPHY
GAMMA CAMERAS
INFORMATION
MATHEMATICAL LOGIC
MATHEMATICAL MODELS
NUMERICAL DATA
TOMOGRAPHY
550601* - Medicine- Unsealed Radionuclides in Diagnostics