| 摘要 |
Incomplete data records owing to an object extent that stretches beyond the scanning field of view (SFOV) constitute a general problem in computed tomography. In these cases, parts of the object are to be reconstructed, for which only incomplete projections from an angular range of less than 180° are available. The application of iterative algorithms such as, for example, the algebraic reconstruction technique (ART) or the simultaneous algebraic reconstruction technique (SART) to this problem of truncated projections cannot lead to a satisfactory solution unless use is made of special boundary conditions. In order to regularize the reconstruction method, in at least one embodiment, information relating to the statistics of the attenuation values of the reconstructed object is also included in the form of the logarithmic probability function of the attenuation values. This information can be obtained from the regions of those image parts which are still completely contained in the SFOV, but nevertheless lie in the vicinity of the region where the object leaves the SFOV. Furthermore, the information can be used in an algebraic reconstruction method by adding a boundary condition term to the cost function to be minimized. Experiments indicate that taking account of this information leads to good estimates with reference to the object in the event of insubstantially truncated projections such as are customary in CT applications, for example.
|
