ArduinoMania

I wrote earlier this week about an applied electronic circuits course for our bioengineering majors  and added some more project ideas in another post.  This weekend, while in southern California for my niece’s wedding, I was given another idea for a project by my sister: a pulse oximeter.

I knew that a pulse oximeter worked by looking at ratios of two different wavelengths of light shining through a finger or earlobe, which have different absorption by oxyhemoglobin and deoxyhemoglobin, but I had never thought about the details.  The first thing that concerned me was how a single ratio of two light intensities could be used—there must be huge variations in how much absorption there is that have nothing to do with oxygenation of hemoglobin.  Rather than puzzle through this on my own, I looked up the Wikipedia article on pulse oximetry.

It turns out that different techniques have been used over the decades.  Some of the early methods used more than two wavelengths of light including some that did not distinguish between the two states of the hemoglobin, in an attempt to quantify the total amount of hemoglobin as well as the ratio of oxy- to deoxy-.  The modern approach, though, is much simpler and more elegant.  The ratio measurements are made repeatedly, and only the fluctuating part of the readings is used.  The arterial blood flow shows a strong pulse signal (at around 0.5–3 Hz), while contaminating signals (hemoglobin in veins, absorption due to tissue other than hemoglobin, fingernail polish, …) are constant. By ignoring the DC signal and using only the fluctuating signal, the oxygen ratios are fairly easily measured.  Note that a pulse is essential to this measurement—continuous flow would not work—hence the name “pulse oximetry”.

There is at least one homebrew pulse oximeter on the web (like this one done by some Polish students and described in Polish, with a Google translation).  As far as I can tell, they used a PWM output from the microprocessor to provide signals alternately to the two LEDs, and to demultiplex the signals from the sensor and do separate analog filtering of the two signals.  It would probably make more sense to reduce the sampling rate to about 100Hz and do all the processing digitally. One thing they did that might be a bit trickier on the Arduino was to adjust the current to the LEDs to keep the DC signal level roughly constant, despite differences in the thickness and opacity of the fingers they were shining the light through.  The Arduino provides PWM outputs, but not real digital-to-analog channels.

They also do not seem to have thought about how to calibrate the device, which could be a problem for our lab also. Since commercial pulse oximeters only cost about $30 these days, we could simply have a couple in the lab for students to compare their results to, making a calibration chart of their reading vs the commercial reading.  (The hard part would be finding variation in oxygenation, since every one in the lab will likely be in the normal range of 95–99% oxygenated.)

The pulse oximeter would be a nice step up from the simple one-channel pulse sensor.  Unfortunately, most of  the design work seems to be on the digital side, rather than the analog side, which may make it a poor fit for the circuits course.