Effect of head model on Monte Carlo modeling of spatial sensitivity distribution for functional near-infrared spectroscopy

Modeling Light propagation within human head to deduce spatial sensitivity distribution (SSD) is important for Near-infrared spectroscopy (NIRS)/imaging (NIRI) and diffuse correlation tomography. Lots of head models have been used on this issue, including layered head model, artificial simplified head model, MRI slices described head model, and visible human head model. Hereinto, visible Chinese human (VCH) head model is considered to be a most faithful presentation of anatomical structure, and has been highlighted to be employed in modeling light propagation. However, it is not practical for all researchers to use VCH head models and actually increasing number of people are using magnet resonance imaging (MRI) head models. Here, all the above head models were simulated and compared, and we focused on the effect of using different head models on predictions of SSD. Our results were in line with the previous reports on the effect of cerebral cortex folding geometry. Moreover, the influence on SSD increases with the fidelity of head models. And surprisingly, the SSD percentages in scalp and gray matter (region of interest) in MRI head model were found to be 80% and 125% higher than in VCH head model. MRI head models induced nonignorable discrepancy in SSD estimation when compared with VCH head model. This study, as we believe, is the first to focus on comparison among full serials of head model on estimating SSD, and provided quantitative evidence for MRI head model users to calibrate their SSD estimation.