Premium
Structured noise in computed tomography: Effects of periodic error sources
Author(s) -
Parker Dennis L.,
Couch John L.,
Peschmann Kristian R.,
Smith Ver
Publication year - 1982
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.595118
Subject(s) - artifact (error) , bessel function , sampling (signal processing) , projection (relational algebra) , periodic function , noise (video) , detector , radius , signal (programming language) , optics , mathematics , scanner , phase (matter) , physics , mathematical analysis , algorithm , computer science , computer vision , image (mathematics) , computer security , quantum mechanics , programming language
The artifact in computed tomography (CT) images due to cyclic projection errors, such as errors due to periodic fluctuations in x‐ray intensity, is derived and verified by computer simulation. Depending upon the relative phase of the error between projections, the artifact is shown to be described by a Bessel function (or functions) of the radial argument which is sinusoidally modulated as a function of angle. Because of the nature of Bessel functions, the artifact is essentially zero up to some minimum radius. When the effects of sampling are neglected, a single fundamental artifact is shown to occur. In fourth‐generation scanners, the fundamental artifact (neglecting sampling) will occur at a detector fan angle of about about 39° (depending on the ratio of the axis‐to‐detector, source‐to‐axis distances). The radius of appearance of this fundamental artifact is independent of the frequency of the periodic error signal and will only be visible in fourth‐generation CT scanners with detector fan angles greater than about 39°. The effects of sampling are derived and illustrated by simulation for first‐, third‐, and fourth‐generation CT‐scanner geometries. It is shown that the effect of sampling is to cause an infinite number of such artifacts to be superimposed in the final image. The radius of appearance of all but the fundamental artifact are shown to be dependent on the frequency of the periodic signal. It is shown that by judicious choice of the sampling parameters relative to the frequency of the periodic error, the artifact can be effectively eliminated.