A comparative study on pre-smoothing techniques for projection data with poisson noise in computed tomography

A computer simulation was performed to compare pre-smoothing techniques for noisy projection data in reconstructed (computed tomographic) images and to find suitable pre-smoothing techniques for reducing the noise in the reconstruction. Some are applied to Poison (signal-dependent) noise, and the others to signal-independent noise obtained by the Anscombe transformation of signal-dependent noise. Selective for signal-dependent noise is found to produce superior results.