Source Detection and Image Reconstruction with Position-Sensitive Gamma-RayDetectors.

Image Reconstruction;Detection;Asymptotic Statistics;Compton Imaging;Electrical Engineering;Engineering;Electrical Engineering: Systems

Lingenfelter, Daniel J.Scott, Clayton D. ;

University of Michigan

关键词: Image Reconstruction;    Detection;    Asymptotic Statistics;    Compton Imaging;    Electrical Engineering;    Engineering;    Electrical Engineering: Systems;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/91441/danling_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Gamma-ray detectors have important applications in security, medicine, and nuclear non-proliferation. This thesis investigates the use of regularization to improve image reconstruction and efficient methods for predicting source detection performance with position-sensitive gamma-ray detectors.Position-sensitive detectors have the ability to measure the spatial gamma emission density around the detector. An image of the spatial gamma emission density where a spatially small source is present will be sparse in the canonical basis, meaning that the emission density is zero for most directions but large for a small number of directions. This work uses regularization to enforce sparsity in the reconstructed image, and proposes a regularizer that effectively enforces sparsity in the reconstructed images.This work also proposes a method for predicting detection performance. Position-sensitive gamma-ray imaging systems are complex and difficult to model both accurately and efficiently. This work investigates the asymptotic properties of tests based on maximum likelihood (ML) estimates under model mismatch, meaning that the statistical model used for detection differs from the true distribution. We propose general expressions for the asymptotic distribution of likelihood-based test statistics when the number of measurements is Poisson. We use the general expressions to derive expressions specific to gamma-ray source detection that one can evaluate using a modest amount of data from a real system or Monte-Carlo simulation. We show empirically with simulated data that the proposed expressions yield more accurate detection performance predictions than expressions that ignore model mismatch. We also use data recorded with a 3D position-sensitive CdZnTe system with a Cs-137 source in a natural background to show that the proposed method is reasonably accurate with real data. These expressions require less data and computation than conventional empirical methods. To quantify the benefit of position-sensitivity, we state and prove a theorem affirming that, asymptotically as scan time becomes large, position-sensitivity increases the area under the receiver operating characteristic curve (AUC) when the background intensity is known, detector sensitivity is spatially uniform, and the system model is correctly specified.