In this paper, a spatial location weighted gradient-based optimization scheme for reducing the computation burden and increasing the reconstruction precision is stated. The method applies to DC diffusionbased optical tomography, where otherwise the reconstruction suffers slow convergence. The inverse approach employs a weighted steepest descent method combined with a conjugate gradient method. A reverse differentiation method is used to efficiently derive the gradient. The reconstruction results confirm that the spatial location weighted optimization method offers a more efficient approach to the DC optical imaging problem than unweighted method does.
© 2002 Optical Society of AmericaPDF Article