A Comparison of Reflectance Pulse Oximetry and near Infrared Spectroscopy for the Determination of the Fraction of Oxygenated Haemoglobin at Low Oxygen Saturations in an Animal Study

Abstract

Near infrared spectroscopy scans of cerebral tissue in vivo were compared to reflectance pulse oximetry of local tissue by comparing each method to in vitro femoral arterial blood measurements in an experimental study on 16 newborn piglets. To test the accuracy, the 16 piglets were desaturated stepwise during general anaesthesia, leading to the following results of linear regression analysis of complete data, linear regression analysis for individual piglets and bias ± precision (1SD). Local oxygen saturation (reflectance pulse oximetry) versus arterial (haemoximeter): linear regression analysis R² = 0.92, n = 207, linear regression analysis for individual piglets: R² = 0.95 ± 0.05*, p < 0.05 versus near infrared spectroscopy, bias ± precision (1SD): 0.15 ± 9.65%; cerebral tissue oxygenation index (near infrared spectroscopy) versus arterial (haemoximeter): linear regression analysis R² = 0.26, n = 161, linear regression analysis for individuals R² = 0.83 ± 0.15, bias ± precision (1SD) −3.33 ± 29%; cerebral concentration of delta oxygenated haemoglobin (near infrared spectroscopy, NIRO 300) versus the corresponding arterial parameter (haemoximeter): linear regression analysis R² = 0.76, n = 189, linear regression analysis for individuals R² = 0.90 ± 0.06, bias ± precision (1SD) −2.54 ± 3 μM cm⁻³. The results of reflectance pulse oximetry in local tissue correlated more closely to in vitro femoral arterial blood measurements (haemoximeter) than results of near infrared spectroscopy, in vivo.