On the convergence of algebraic reconstruction methods for computer tomography

This paper discusses the convergence of algebraic reconstruction methods in computer tomography. They require high computing capacity because of their iterative nature. With increasing computational power the algebraic reconstruction methods gain more influence. The most obvious problem of algebraic methods is the fact that reconstruction quality is affected by distortions increasing with iteration numbers. This is some times interpreted as a bad convergence behaviour of the algorithm. This paper explains the nature of these distortions and demonstrates in contrary a good convergence of algebraic reconstruction methods.