BACKGROUND: Reflectance near-infrared spectroscopy (600-2200 nm) can noninvasively probe deep into tissues. Blood is the predominant absorber of near-infrared light in biological tissues. We investigated the feasibility of using reflectance near-infrared spectroscopy to measure blood pH in vitro. METHODS: Reflectance near-infrared spectra (600-2200 nm) were obtained with a fiberoptic probe immersed in diluted human packed red blood cells maintained at 37 degrees C. Changes in pH (6.800-7.600) were induced by: (1) varying the partial pressure of carbon dioxide by the bubbling of mixtures of humidified carbon dioxide and nitrogen gas through the blood; and (2) adding 1 N HCl/NaOH. Humidified oxygen gas was bubbled through the blood to generate variations in oxygen saturation. After each titration of pH, the spectrum was recorded and blood was sampled for the measurement of: pH, pCO(2), and pO(2) using blood gas analysis; and hemoglobin concentration and oxygen saturation using co-oximetry. Samples from three separate pH titrations were combined (120 total samples) and analyzed using partial least-squares analysis to generate a mathematical model relating spectral changes to pH (calibration set). This model was then used to predict the pH of a set of 36 pH titrations (prediction set). RESULTS: Quantitative and qualitiative analyses of the spectra in the calibration set found that spectral changes in the wavelength range, 650-1050 nm, were directly related to changes in pH. First-derivative-treated spectra from the calibration set, analyzed using partial least-squares analysis, generated a mathematical model with a cross-validated r(2) of 0.939 and a standard error of calibration of 0.046 pH unit. When this model was applied to the prediction set, with an offset correction, the r(2) was 0.936 with a standard error of prediction of 0.050 pH unit. CONCLUSION: Blood pH can be predicted in vitro with clinical significance using reflectance near-infrared spectroscopy (650-1050 nm) within a standard error of 0.050 pH unit.
BACKGROUND: Reflectance near-infrared spectroscopy (600-2200 nm) can noninvasively probe deep into tissues. Blood is the predominant absorber of near-infrared light in biological tissues. We investigated the feasibility of using reflectance near-infrared spectroscopy to measure blood pH in vitro. METHODS: Reflectance near-infrared spectra (600-2200 nm) were obtained with a fiberoptic probe immersed in diluted human packed red blood cells maintained at 37 degrees C. Changes in pH (6.800-7.600) were induced by: (1) varying the partial pressure of carbon dioxide by the bubbling of mixtures of humidified carbon dioxide and nitrogen gas through the blood; and (2) adding 1 N HCl/NaOH. Humidified oxygen gas was bubbled through the blood to generate variations in oxygen saturation. After each titration of pH, the spectrum was recorded and blood was sampled for the measurement of: pH, pCO(2), and pO(2) using blood gas analysis; and hemoglobin concentration and oxygen saturation using co-oximetry. Samples from three separate pH titrations were combined (120 total samples) and analyzed using partial least-squares analysis to generate a mathematical model relating spectral changes to pH (calibration set). This model was then used to predict the pH of a set of 36 pH titrations (prediction set). RESULTS: Quantitative and qualitiative analyses of the spectra in the calibration set found that spectral changes in the wavelength range, 650-1050 nm, were directly related to changes in pH. First-derivative-treated spectra from the calibration set, analyzed using partial least-squares analysis, generated a mathematical model with a cross-validated r(2) of 0.939 and a standard error of calibration of 0.046 pH unit. When this model was applied to the prediction set, with an offset correction, the r(2) was 0.936 with a standard error of prediction of 0.050 pH unit. CONCLUSION: Blood pH can be predicted in vitro with clinical significance using reflectance near-infrared spectroscopy (650-1050 nm) within a standard error of 0.050 pH unit.