Image restoration in computed tomography: the spatially invariant point spread function.

IEEE transactions on medical imaging

PubMedID: 18222894

Rathee S, Koles ZJ, Overton TR. Image restoration in computed tomography: the spatially invariant point spread function. IEEE Trans Med Imaging. 2008;11(4):530-8.
Image restoration to deblur smoothing caused by the finite-size X-ray beam profile for a simulated computed tomography (CT) system is presented. Three simple image restoration methods are compared when the point-spread-function (PSF) is spatially invariant. In the first restoration method, an iterative least squares solution, regularized with the image norm and constrained by the boundary of the object, is obtained from the projection data. In the second method, a Wiener filter, designed using the power spectrum of CT noise, is applied to the reconstructed CT image. The third method obtains a weighted least-squares solution, by iteration, from the reconstructed CT image; the solution is regularized with the weighted image norm. Restored images were compared with the image obtained using filtered backprojection method. Differences between these images were evaluated qualitatively.