The larger the number of projections applied on the original image, the more accurate the reconstructed image becomes. In computed tomography, many projections of the object are first generated from different angles. Then filtered back-projections are applied to reconstruct a 2D image of the structure of a particular cross section of the image. This is the basic idea used in X-ray medical imaging.
The Radon transform and the inverse Radon transform (both added in Mathematica
8) are used to simulate this method. Up to 128 projections can be taken between
. Then applying the inverse Radon transform on the resulting image gives the filtered back-projection image.
In this Demonstration, only one filtered back projection is used per projection. The magnitude spectrum of the reconstructed image (the inverse Radon image) is updated as more back projections are applied, showing that the spectrum approaches that of the original image. Ram–Lak and cosine ramp filters for inverse Radon transform generate the clearest reconstruction; however, streak lines appear across the reconstructed image; these do not appear in some of the other filters nor for the nonfiltered image.
The "n" slider represents the number of projections or angles to apply. The parameter
option for the inverse Radon. You can adjust the 2D frequency spectrum of the images for better viewing. A checkbox lets you change the view of the image magnitude spectrum from 2D to 3D. For original color images, a checkbox can be used to process the image in gray only, as processing the image in color requires more time and memory.
 A. C. Kak and M. Slaney, Principles of Computerized Tomographic Imaging
, New Brunswick, NJ: IEEE Press, 1988.
 H. Murrell, "Computer-Aided Tomography," The Mathematica Journal
(2), 1996 pp. 60–65.