I'm designing a spectral-differencing sensor for use in high speed egg-grading machines, to detect blood spots in eggs. This is an interesting problem because it's dominated by nuisance data. Nuisance data are sample variations that produce a signal but aren't what we care about measuring. In this case, porphyrins in brown eggshells have strong absorption features that overlap those of haemoglobin.
I've ordered a small CCD spectrometer to measure a bunch of eggs and a bunch of optical sources.
Since eggs are such a basic foodstuff, there's a fair amount of literature on this, but the net is that there's an accessible haemoglobin absorption feature near 577 nm that can be distinguished from absorption on the red side due to protoporphyrin (the brown-egg pigment) and the normal high absorption of all eggs on the blue side. It's an interesting problem because of the ancillary constraints: it has to be easy to clean and disinfect, has to work in room lights, and has to cope with a 1000:1 absorption range (3 absorption units). Oh, and of course it has to be accurate, reliable, inexpensive, and easy to service.