New developments in frequency domain optical tomography. Part II: Application with a L-BFGS associated to an inexact line search

Abstract Gradient-based algorithms in optical tomography have shown efficiency and robustness due to their self-regularization effect, as they track the largest errors first, which tends to stabilize the reconstruction scheme. This second part shows the application of the presented developments in the first part for the reconstruction of optical properties in complex geometry while taking into account the collimated source direction with such an algorithm. The reconstruction scheme is based on the limited memory BFGS type associated to an inexact line search in order to avoid numerous evaluations of the objective function. Normalization of the objective function with measurements and independent scaling of its gradient are used to improve the quality of the reconstruction. The results show that the algorithm is efficient compared to other solvers with a better recovering of both the absorption and scattering coefficients.