Diffuse reflectance spectrophotometry with visible light: comparison of four different methods in a tissue phantom

The purpose of the study was to compare algorithms of four methods (plus two modifications) for spectrophotometric haemoglobin saturation measurements. Comparison was made in tissue phantoms basically consisting of a phosphate buffer, Intralipid and blood, allowing samples to be taken for reference measurements. Three experimental series were made. In experiment A (eight phantoms) we used the Knoefel method and measured specific extinction coefficients with a reflection spectrophotometer. In experiment B (six phantoms) the fully oxygenated phantoms were gradually deoxygenated with baker's yeast, and simultaneous measurements were made with our spectrophotometer and with a reference oxymeter (ABL-605) in 3 min intervals. For each spectrophotometric measurement haemoglobin saturation was calculated with all algorithms and modifications, and compared with reference. In experiment C (11 phantoms) we evaluated the ability of a modification of the Knoefel method to measure haemoglobin concentration in absolute quantities using extinction coefficients from experiment A.Results. Experiment A: with the Knoefel method extinction coefficients (+/-SD) for oxyhaemoglobin at 553.04 and 573.75 nm were 1.117 (+/-0.0396) ODmM(-1) and 1.680 (+/- 0.0815) ODmM(-1), respectively, and for deoxyhaemoglobin 1.205 (+/- 0.0514) ODmM(-1) and 0.953 (+/-0.0487) ODmM(-1), respectively. Experiment B: high correlation with the reference was found in all methods (r = 0.94-0.97). However, agreement varied from evidently wrong in method 3 and the original method 4 (e.g. saturation above 160%) to high agreement in method 2 as well as the modifications of methods 1 and 4, where oxygen dissociation curves were close to the reference method. Experiment C: with the modified Knoefel method the mean haemoglobin concentration difference from reference was 8.3% and the correlation was high (r = 0.91). We conclude that method 2 and the modifications of 1 and 4 were superior to the others, but depended on known values in the same or similar phantoms. The original method 1 was independent of results from the tissue phantoms, but agreement was slightly poorer. Method 3 and the original method 4 could not be recommended. The ability of the modified method 1 to measure haemoglobin concentration is promising, but needs further development.