A magnetic resonance imaging (MRI) guided approach is introduced aimed at improving the spatial resolution and image localization errors of diffuse optical tomography (DOT). This approach allows incorporation of the priori MRI-based anatomical information through a segmented finite element mesh of a human brain. Since there are few existing literatures discussing the classification choice of the brain model and the settings of optical parameters, we propose three imaging quality evaluation indices in 3D space using point-spread-function (PSF) simulation, and provide a thorough evaluation of the image quality under condition of different tissue the classification and optical parameters. Meanwhile, reconstructed images are given in 3D space with depth information. It is indicated that a four-layer brain model: scalp, skull, brain and cerebrospinal fluid, produces the best image quality and the image quality is relatively even over the entire field of view. The effect of optical parameters′ variation within a certain range is relatively small and negligible.