Computer Tomography is a technique used to reconstruct a cross-section image of an object from its line-integral projections. A complete set of these projections is also known as a sinogram. Conventional image reconstruction algorithms implemented on existing CT systems require the collection of projection data covering the whole measurement range. Unfortunately, in many practical situations, it is not always possible to obtain all the projection data. This occurs, for example, if the tissue distributions to be measured contain X-ray opaque objects (metallic implants, screws, or clips) which attenuate the rays completely; the detector readings for these parts will be missing. This is the so-called hollow projection problem. If conventional algorithms without any restoration techniques a...[ Read more ]

Computer Tomography is a technique used to reconstruct a cross-section image of an object from its line-integral projections. A complete set of these projections is also known as a sinogram. Conventional image reconstruction algorithms implemented on existing CT systems require the collection of projection data covering the whole measurement range. Unfortunately, in many practical situations, it is not always possible to obtain all the projection data. This occurs, for example, if the tissue distributions to be measured contain X-ray opaque objects (metallic implants, screws, or clips) which attenuate the rays completely; the detector readings for these parts will be missing. This is the so-called hollow projection problem. If conventional algorithms without any restoration techniques are used in these limited-data situations, the resultant images suffer from severe streak artifacts.

In this thesis we propose an interpolation method to restore a sinogram from its available incomplete sinogram. This algorithm is based on the Ludwig- Helgason Consistency Theorem and uses the edge values of the missing part of the incomplete sinogram. In our algorithm the missing range of projection values can be arbitrary and no a priori knowledge is needed. A computationally efficient algorithm was developed to solve the interpolation problem. We also assessed the effectiveness of our algorithm performed on noisy sinograms. Finally, by integrating with other restoration techniques, a hybrid algorithm was developed for fine tuning the restored image. Computer simulation results demonstrate the efficiency of the proposed method.