Influence of the Signal-To-Noise Ratio on Variance of Chromophore Concentration Quantification in Broadband Near-Infrared Spectroscopy

Abstract

This study presented a theoretical or analytical approach to quantify how the signal-to-noise ratio (SNR) of a near infrared spectroscopy (NIRS) device influences the accuracy on calculated changes of oxy-hemoglobin (Δ[HbO]), deoxy-hemoglobin (Δ[HHb]), and oxidized cytochrome c oxidase (Δ[oxCCO]). In theory, all NIRS experimental measurements include variations due to thermal or electrical noise, drifts, and disturbance of the device. Since the computed concentration results are highly associated with device-driven variations, in this study, we applied the error propagation analysis to compute the variability or variance of Δ[HbO], Δ[HHb], and Δ[oxCCO] depending on the system SNR. The quantitative expressions of variance or standard deviations of changes in chromophore concentrations were derived based on the error propagation analysis and the modified Beer-Lambert law. In order to compare and confirm the derived variances versus those from the actual measurements, we conducted two sets of broadband NIRS (bbNIRS) measurements using a solid tissue phantom and the human forearm. A Monte Carlo framework was also executed to simulate the bbNIRS data under two physiological conditions for further confirmation of the theoretical analysis. Finally, the confirmed expression for error propagation was utilized for quantitative analyses to guide optimal selections of wavelength ranges and different wavelength combinations for minimal variances of Δ[HbO], Δ[HHb], and Δ[oxCCO] in actual experiments.