Linear analysis of single-shot dual-energy computed tomography with a multilayer detector

Abstract

The sandwich-like multilayer detector can have a unique characteristic when it employs two detector layers having different spatial-resolution properties to each other. Reconstruction of dual-energy (DE) images using a weighted log-subtraction technique then results in the unsharp masking-like effect on the resultant images. We have developed a simple model describing the signal and noise in the sandwich detector-based DE computed tomography (CT) using a linear-systems theory. Applying the sandwich detector to a laboratory micro-CT system, we measured the Fourier metrics, such as modulation-transfer function, noise-power spectrum, and noise-equivalent number of quanta (NEQ), using thin wire and water phantoms. The agreements between the measured Fourier metrics and the theoretical models were excellent. The DE-CT for a postmortem mouse showed an enhanced details of bone structures, and the theoretical model explained the reason as showing a shift of the peak position of NEQ to high-frequency regions.