Minimum spectral resolution for continuous-wave hyperspectral near-infrared tissue spectroscopy.

Abstract

Significance: Continuous-wave hyperspectral near-infrared spectroscopy (h-NIRS) is a reliable and cost-effective technique for noninvasive monitoring of tissue blood content and oxygenation at the point-of-care; however, most h-NIRS devices are based on expensive custom-built spectrometers. For widespread adoption, low-cost, miniature, off-the-shelf spectrometers are needed. To guide the development of such spectrometers, a standard for spectral resolution must first be defined.

Aim: We aim to identify the minimum spectral resolution needed for h-NIRS devices to accurately measure oxy- and deoxy-hemoglobin (HbO and Hb) concentrations in tissue.

Approach: h-NIRS measurements were acquired from fully oxygenated and deoxygenated blood-lipid phantoms at 13 spectral resolutions. Data for other oxygenation levels were simulated using NIRFAST. HbO and Hb concentrations were estimated at each resolution and compared with the ground truth hemoglobin concentration.

Results: The concentration of Hb was estimated with high accuracy for resolutions up to 10 nm, whereas HbO estimates were more variable. For both chromophores, the accuracy of the estimation gradually decreased with resolutions beyond 10 nm.

Conclusions: Spectral resolutions up to 10 nm can be used for h-NIRS without compromising the accuracy of estimating tissue blood content and oxygenation.