The study aims to investigate and test a new algorithm applying to the sampled projection to generate high quality images by reducing the noise in the cone-beam computed tomography (CBCT). A single sampling without object is first employed in scanning to obtain the noise as the knowledge, then a random Gaussian Matrix (GM) is used to get a new noise map that is consistent with the previous one based on the knowledge. Study results demonstrated significant improvement in the signal-to-noise ratios (SNRs) of the images by overlapping the noise map on the projection for average. Since the noise is reduced, it has the potential to improve projection quality by using the original noise sampled on flat panel detector (FDP).
© 2018 The Author(s)PDF Article