Raman Spectroscopy of Blood and Urine Specimens - Emerging Raman Applications and Techniques in Biomedical and Pharmaceutical Fields - page 396

16.5.4 Urine
As noted above, concentrations of native urine chemicals hold less clinical
importance than those from blood, and there has been correspondingly less
research devoted to this area. Early papers testing the feasibility of measuring
certain chemicals in urine were published by Premasiri et al. [23] and Dou et al.
[24], though validated predictions of concentrations at native levels in actual
urine specimens were not carried out. Improving upon preliminary work by
McMurdy that used a Cassegrainian sampling geometry [25], Qi et al. used the
same LCOF system as in their blood serum work to analyze 69 in vitro urine
specimens [5]. They obtained strong PLS-based correlations for creatinine and
urea content, limited by the accuracy of the reference chemistry in both cases.
The creatinine error was 0.38 mM, empirically in line with the sorts of results
found in blood serum, although unlike those cases this was a limit imposed
by the reference chemistry, not the Raman methodology.
Some particular applications of Raman-based urine analysis have recently
been reported. Park et al. explored the ability of Raman spectroscopy to
detect abnormally high levels of glucose in diluted urine, with the goal of
installing such technology on toilet bowls to monitor diabetic users [26, 27].
Using a linear discrimination technique, they reported an ability to discrim-
inate between spectra of pure water and water with 0.28 mM glucose. They
claimed that if they could reproduce these results in urine and at somewhat
lower concentration, then they would be able to distinguish between normal
and elevated levels in urine that has been diluted, presumably by the toilet
bowl water. The possibility of modifying the toilet bowl to enable capture and
analysis of undiluted urine was not mentioned. Meanwhile, Guimaraes et al.
doped the drug ephedrine into urine specimens and also obtained the urine
of an ephedrine-injected rat. By simple calculation of a peak area for a band
at 1002 cm − 1 corresponding to the drug, they report an ability to detect drug
concentrations as low as 0.03 mM [28], which is plausible from first principles
(especially since the drug probably has a stronger Raman cross section than
most natively occurring chemical species). Their regression line does not ex-
trapolate to zero Raman signal for zero drug concentration, probably because
the urea Raman band at 1005 cm − 1 is not perfectly removed, so the lower
detection limit claimed is slightly suspect.
16.6 Conclusion
Spontaneous Raman spectroscopy has the ability to provide clinically relevant
chemical concentration measurements of multiple analytes in biofluids. Blood
serum, whole blood, and urine have all been studied. The detection limit (as-
suming a few hundred seconds of spectral acquisition) appears, based upon
fundamental noise considerations, to be around 0.1 mM for most biochemicals;
this places several important analytes within reach but certainly precludes