Recursive least squares image reconstruction

A recursive least squares algorithm is formulated in the context of image reconstruction, using computed tomography (CT) as an example. Compared to traditional algebraic reconstruction techniques (ART), the new algorithm converges in as little as one iteration and has the ability to incorporate a priori information such as image smoothness. Several approximations of the algorithm are proposed and it is demonstrated that they can be implemented for practical image reconstruction problems and obtain very good results.