PET image reconstruction with allowance for errors in the system model

In tomographic image reconstruction, it is usually assumed that the system matrix (describing the point source response) is known exactly. Here, the authors attempt to model the system matrix more realistically as the sum of a known part and an unknown random part that accounts for any errors. By using some simplifying assumptions, the authors develop a penalized weighted least squares (PWLS) reconstruction algorithm for this problem. Their experiments indicate that this approach can, indeed, lead to significant improvements in the reconstructed image, both visually and quantitatively.