Sensitivity of functional near-infrared spectroscopy to optical properties in brain imaging

Abstract

Functional near-infrared spectroscopy (fNIRS) is a noninvasive optical technique for assessing spatial brain activation by determining the relative changes in the concentrations of oxy- and deoxyhemoglobin (HbO and HbR) in different regions of the brain. The modified Beer-Lambert Law (mBLL) is essential for calculating the relative concentrations of HbO and HbR, and the differential pathlength factor (DPF) is a key component of the mBLL. In this paper, we investigate how error in DPF estimates translates to relative concentration calculations for HbO and HbR. To do this, we generate various amounts of error in DPF values and calculated the error in the resulting concentration data. We then use a two-regression fit to generate a model of concentration error as a function of DPF error. We also compare different reported DPF values to assess their impact on concentration error and to verify our findings with the concentration-error model. We find that age appears to be a greater contributing factor to concentration error than same-subject anatomical differences, and we also find that DPF over- and underestimation in equal amounts will produce unequal amounts of concentration error.