Noise and contrast detection in computed tomography images

Abstract

A discrete representation of the reconstruction process is used in an analysis of noise in computed tomography (CT) images. This model is consistent with the method of data collection in actual machines. An expression is derived which predicts the variance on the measured linear attenuation coefficient of a single pixel in an image. The dependence of the variance on various CT scanner design parameters such as pixel size, slice width, scan time, number of detectors, etc., is then described. The variation of noise with sampling area is theoretically explained. These predictions are in good agreement with a set of experimental measurements made on a range of CT scanners. The equivalent sampling aperture of the CT process is determined and the effect of the reconstruction filter on the variance of the linear attenuation coefficient is also noted, in particular, the choice and its consequences for reconstructed images and noise behaviour. The theory has been extended to include contrast detail behaviour, and these predictions compare favourably with experimental measurements. The theory predicts that image smoothing will have little effect on the contrast-detail detectability behaviour of reconstructed images.

In vivosmall-animal imaging using micro-CT and digital subtraction angiography

“…The dose is a function of the energy deposited through photon absorption and energy loss via Compton scattering interactions. The smaller voxels of micro-CT (relative to clinical CT) require much higher dose, since any voxel – independent of its size – needs to interact with a certain number of x-ray photons for adequate image quality (Faulkner and Moores, 1984; Brooks and Di Chiro, 1976). Therefore, in order to compensate for the reduction of the voxel size, the photon fluence must be much larger in micro-CT to achieve a signal-to-noise ratio comparable to clinical CT.…”

Image reconstruction from limited angle projections collected by multisource interior x-ray imaging systems

“…In particular, acquiring CT images of the animal heart with organ definition and temporal resolution comparable to that in humans requires ~10-fold improvement in temporal resolution, and ~100-fold increment in spatial resolution. On the other hand, the high spatial resolution requires much higher dose relative to clinical CT, because any voxel must interact with a decent number of x-ray photons for adequate image quality (Faulkner and Moores, 1984, Brooks and Chiro, 1976). Therefore, to compensate for the reduction in focal spot and image voxel size, the detector integration time and scanning speed have become too long to be practical in many important scenarios.…”

Fourier Analysis of Noise Characteristics in Cone-Beam Microtomography Laboratory Scanners

“…Limiting the scope to CT, Riederer et al [27] derived an analytical CT NPS formalism and showed it to be proportional to the squared transfer function of the reconstruction filter. Furthermore, Wagner et al [28], [29], Hanson [30], and Faulkner and Moores [31] connected the analytical NPS to NEQ and applied the NEQ to simple detection tasks. On the other hand, Kijewski and Judy [32] showed an asymmetric CT NPS due to aliasing using a discrete analysis.…”