Monte Carlo Simulation of light propagation in the adult brain

摘要

When near-infrared spectroscopy (NIRS) is applied noninvasively to the adult head for brain monitoring, extra-cerebral bone and surface tissue exert a substantial influence on the cerebral signal. Most attempts to subtract extra-cerebral contamination involve spatially resolved spectroscopy (SRS). However, inter-individual variability of anatomy restrict the reliability of SRS. We simulated the light propagation with Monte Carlo techniques on the basis of anatomical structures determined from 3D-magnetic resonance imaging (MRI) exhibiting a voxel resolution of 0.8 x 0.8 x 0.8mm~3 for three different pairs of T1/T2 values each. The MRI data were used to define the material light absorption and dispersion coefficients for each voxel. The resulting spatial matrix was applied in the Monte Carlo Simulation to determine the light propagation in the cerebral cortex and overlaying structures. The accuracy of the Monte Carlo Simulation was furthermore increased by using a constant optical pathlength for the photons which was less than the median optical pathlength of the different materials. Based on our simulations we found a differential pathlength factor (DPF) of 6.15 which is close to the value of 5.9 found in the literature for a distance of 4.5cm between the external sensors. Furthermore, we weighted the spatial probability distribution of the photons within the different tissues with the probabilities of the relative blood volume within the tissue. The results show that 50% of the NIRS signal is determined by the grey matter of the cerebral cortex which allows us to conclude that NIRS can produce meanigful cerebral blood flow measurements providing that the necessary corrections for extracerebral contamination are included.