A new reconstruction methodology for computerized tomography: FRECT (Fourier regularized computed tomography)

Presents a new regularization method for tomography, which combines features from the FBP algorithm and regularization theory. The preliminary results shown here are extremely encouraging, and seem to indicate that a substantial improvement of the stability of the reconstruction and of quality of the image can be expected, especially in situation of sparse angular sampling. One learns from the Fourier Slice theorem that constraints obtained from the measurements can be read off the Fourier domain. The regularization constraints should act beyond the experimental area, so as to avoid conflicts between these two types of constraints. In addition, such regularization constraints can clearly be interpreted in terms of filtering or Point Spread Function, which makes the reconstructed images more legible. In order to account for the full complexity of the physical relationship between the distribution of interest and the data acquired by the camera, the method needs to refined. This is currently under study, and will form the subject of further publications