T1 weighted magnetic resonance imaging of the brain was segmented into scalp, skull, cerebrospinal fluid, gray matter and white matter, and then meshed in volume with different tissues. The volumetric mesh based on subject specific anatomy was used to build forward optical model and reconstruct image in the diffuse optical tomography for solving low position-accuracy, ill-condition and under-determined problems caused by hemisphere model or brain atlas. In the visual experiments of counter-clockwise wedge stimulus and expanding ring stimulus, the diffuse optical tomography guided by magnetic resonance imaging can obtain higher space resolution and more accurate position for activated oxygenated haemoglobin signals.