Application of the approximate inverse to 3D \(X\)-ray CT and ultrasound tomography (Q4356949)

We study operator equations \(Af=g\) for operators between Hilbert spaces \(X\) and \(Y\). Both the cases of linear operators \(A\) and of nonlinear \(A\) with a special structure are treated. Approximate inverse means a solution operator which maps the data \(g\) to a stable approximation of the solution of the ill-posed problem \(Af=g\). This inversion operator is precomputed without using the data \(g\). We show that the approximate inverse is a regularization method. For nonlinear operators we combine these ideas with the results of \textit{R. Snieder} [(*) Inverse Probl. 7, No. 3, 409-433 (1991; Zbl 0737.35155)] who generalized the Backus-Gilbert method to nonlinear problems. A detailed analysis of the method shows that the computational effort is much smaller than in (*). Applications to 3D \(x\)-ray computerized tomography (CT) and ultrasound CT conclude the paper. For the entire collection see [Zbl 0871.00041].