Evaluation of multi-wavelength derivative spectra for quantitative applications in clinical chemistry

We have used the quantitation of methemoglobin in the presence of a light-scattering suspension as a model to evaluate and compare the relative merits of zeroth-, first-, and second-derivative spectroscopy for problems in clinical chemistry. Data for second derivatives are most effective in nullifying the effect of the background spectrum; data for first derivatives are less effective; and data for absorbance (zeroth derivative) are least effective. In one set of experiments, variable amounts of the light-scattering component were added to a methemoglobin solution to increase the apparent absorbance up to about 230% of the absorbance of the original solution. Whereas computation with data for absorbance would have yielded errors approaching 100%, concentrations computed with second-derivative spectral data yielded a systematic error (bias) of only 0.5% and a relative standard deviation (CV) of 1.6%.