Simon and I have just got signed up to build a trans-abdominal fetal blood oxygenation sensor--i.e. a pulse-ox for the baby that works from outside the mom's abdomen.
The current method uses ultrasound to look at how responsive the baby's heart rate is to the mom's contractions, in order to detect fetal respiratory distress.
It's a very sensitive measurement, but unfortunately it's also highly sensitive to a lot of other things that have nothing to do with fetal distress. That leads to a lot of unnecessary caesarean sections, and hence to a lot of harm to the patients.
This one, which we're doing in cooperation with a start-up spun off from UC Davis, does the usual pulse-ox thing with LEDs and photodiodes sensing haemoglobin absorption modulated by the heart rate. Since the mom's pulse is usually slower than the child's, in principle this is unambiguous.
The issue as usual is getting enough photons from the right depth in the tissue, and then not corrupting them with all sorts of circuit noise. In a highly-scattering medium like tissue, the light intensity distribution tends to follow Laplace's equation, so it falls off steeply with depth. For the same reason, you want to separate the source and detector by approximately the desired sensing depth--to get 100 mm into the tissue, you want them 100 mm apart.
I'm expecting low picoamps of signal even with big fat photodiodes. That'll be okay for the proof-of-concept version, but after they get an external funding round they'll want to make the probe disposable, meaning cheeeeep, like AliExpress cheap.
Doing amazing things(*) with nearly zero apparatus is my favourite sort of project.
Fun stuff.
Cheers
Phil Hobbs
(*) For sufficiently-permissive definitions of 'amazing', of course. ;)