Diffuse optical tomography is a non-invasive and non-ionizing optical imaging technique with low cost, while it suffers from low spatial resolution and is very difficult to achieve quantitative measurement. In order to improve the resolution and reconstruct the optical coefficients accurately, in this paper, we present an image reconstruction algorithm based on finite element method for steady-state diffuse tomography with structural priori information. Imaging model is characterized by the steady-state diffuse equation. The spatial structural information from micro-CT is introduced into the inverse problem by the Laplace regularization and Levenberg-Marquardt method to solve the inverse problem where the Jacobian matrix is obtained by adjoint method. The simulation results show that the algorithm presented is able to obtain the accurate distribution of optical coefficients and increase the convergence speed of iteration evidently.