A correction procedure for the self-absorption artifacts in X-ray Compton tomography

Transmission tomography is a powerful tool in several fields of investigation, ranging from medicine to industrial quality control. It is based on the radiation that doesn't interact with the sample and that reaches the detector. However, this kind of technique cannot always be used or the information from it is poor. In some of these cases other kinds of tomography can be applied such as scattering (Compton or Rayleigh) tomography or fluorescence tomography. The use of the latter techniques requires a different reconstruction technique of that used in trasmission tomography. Some adaptation of the backprojection algorithm is often used, but in this case a complete scan of the sample is required with a fixed geometry. Another type of reconstruction algorithms are the iterative ones, but they present a high computational cost and so only small, low spatial resolution matrix, can be treated. In the past another kind of scanning technique has been proposed. It is based on two linear translations (x-y) against translation and rotation required by the backprojection algorithm. The x-y scanning presents a big advantage compared with the backprojection scanning, i.e. the reconstruction does not need to be performed on the whole sample; it can also be performed on a part of the sample, the region of interest (ROI). However this scanning technique is affected by a big drawback: the influence of X-Ray self-absortption by the matrix is stronger than the backprojection scanning producing a lower quality of the reconstruction. Several correction techniques have been described in the literature but they require an a-priori knowledge of the matrix composition or a dual-energy scanning. In this paper a recently developed technique is described. It is based only on the assumption that the Compton effect is predominant (this condition is usually found in the Compton tomography experiments described in the literature) with respect to the other interaction effect. If it is true the algorithm will be able to perfectly correct the self-absorpion effect, even when the quality of the spectra, i.e., the statistics, is poor. Some results are reported and discussed.