Regularized Equally Sloped Tomography Algorithm for Low Dose X-Ray Computed Tomography

X-ray Computed Tomography is widely used in a broad range of fields including astronomy, geophysics, material science, biology and radiology. It has the advantage to achieve the cross section information of the object without physically damaging or penetrating it, which is especially important for in-vivo biology imaging systems or clinical imaging systems. However, due to the requirement of sufficiently high flux projections from multiple directions for achieving high quality images, a major concern in X-ray CT is the unavoidable radiation dose delivered to the imaging objects, especially to the more radiosensitive patients or biology specimens. With the conventional unregularized tomography algorithms, the accuracy of the reconstruction images has to be compromised by lowering the number of projections or reducing the source flux per projection. Therefore, many more sophisticated reconstruction algorithms have been developed recently to solve the missing data and suppress the noise level in the low dose tomography modalities by incorporating mathematical regularization techniques into a real space iterative process. These methods perform well under certain circumstances but also have the limitation in computational speed, which is crucial for real time imaging systems. In this dissertation, an efficient iterative Fourier based reconstruct technique termed Equally- Sloped Tomography (EST), which combines the efficiency of Fourier transform and advanced iterative process and allows for accurate tomographic reconstruction from low flux and undersampled projection data, is presented and investigated. This work focuses on integrating the mathematical regularization methods into the EST algorithm, studying its performances through numerical experiments, comparing the results with other reconstruction algorithms and developing the data preparation procedure for three major tomography geometries: parallel beam, fan beam and helical cone beam. Furthermore, the algorithm was implemented into two important tomography systems: the X-ray medical CT and phase contrast X-ray mammography CT. The performances were studied by conducting experiments using phantom and clinical data at different low dose levels. After carefully evaluated by both visualized comparisons and quantified measurements, the results demonstrate that the regularized EST algorithm is capable to computational efficiently achieve significant radiation dose reduction through both the reduction of flux and the number of projections, while producing comparable image quality results as the full-dose conventional reconstructions do