Co-oximetry

Co-oximetry determines hemoglobin composition using optical measurements at multiple wavelengths. In a laboratory blood analyzer, it can distinguish oxyhemoglobin, deoxyhemoglobin, carboxyhemoglobin and methemoglobin rather than infer all hemoglobin behavior from a single oxygen-related number. A manufacturer description explicitly identifies direct optical spectrophotometry as the measurement method for its analyzer's co-oximetry parameters.^[ref]

An idealized mixture model makes the separating information explicit:where $A_j$ is measured absorbance, $\ell$ the path length, $c_i$ a species concentration, ${\epsilon }_i$ its molar absorption coefficient and $b_j$ a background term. Estimating four unconstrained concentrations requires sufficient independent spectral observations and known response coefficients; merely having four wavelength labels does not guarantee independence. An isosbestic point shared by two species supplies information about their combined amount but cannot by itself separate their proportions.

The denominator of a reported fraction matters. In a simplified four-form example, stipulate $90\%$ oxyhemoglobin, $5\%$ deoxyhemoglobin, $4\%$ carboxyhemoglobin and $1\%$ methemoglobin, all expressed on the same total-hemoglobin basis. Fractional oxyhemoglobin is $0.90$, but the functional oxygen saturation isThese differ because the second denominator excludes the other two forms. The manufacturer's parameter description distinguishes these definitions.^[ref] The chosen fractions are an explanatory example, not clinical thresholds.

A blood-sample optical assay and a noninvasive pulsatile optical measurement use different observation models. Their output labels and denominators must be established from the specific method, rather than inferred from the common word oximetry. Likewise, oxygen partial pressure describes dissolved oxygen equilibrium and is not itself a measurement of all hemoglobin fractions.

An instrument test can compare reference mixtures with independently established composition, record full spectral residuals and vary a suspected interferent. Agreement of one derived saturation number alone does not validate every inferred species concentration. Background, spectral overlap and sample handling belong in the measurement model and its uncertainty.