Statistical methods for reconstruction of parametric images from dynamic PET data

Mustafa E Kamasak, Purdue University

Abstract

In this thesis, we demonstrate and evaluate a new and efficient method for direct voxel-wise reconstruction of kinetic parameter images using all frames of the PET data. The direct parametric image reconstruction is formulated in a Bayesian framework, and uses the parametric iterative coordinate descent (PICD) algorithm to solve the resulting optimization problem. The PICD algorithm is computationally efficient and is implemented with spatial regularization in the domain of the physiologically relevant parameters. Our experimental simulations of a rat head imaged in a working small animal scanner indicate that direct parametric reconstruction can substantially reduce estimation error in kinetic parameters, as compared to indirect methods, without appreciably increasing computation. The PICD algorithm is also used to reconstruct the parametric images of a monkey brain directly from the EXACT HR+ sinograms. PICD reconstructions are compared with the estimates using image domain methods. The results have shown that the proposed direct reconstruction algorithm can produce higher resolution parametric reconstructions. We simultaneously estimate both the kinetic parameters at each voxel and the model-based plasma input function directly from the sinogram data. The plasma model parameters are initialized with an image domain method to avoid local minima, and multiresolution optimization is used to perform the required reconstruction. Good initial guesses for the plasma parameters are required for the algorithm to converge to the correct answer. This method can estimate some of the kinetic parameters (k2, k3, k4, BP), but it can only estimate the others (k1, VD) within a scale factor.

Degree

Ph.D.

Advisors

Bouman, Purdue University.

Subject Area

Electrical engineering|Biomedical research

COinS