Oxygen Saturation - SpO2 Measurement

In the recent past I have seen a lot of questions from fellow community members about how to calculate Oxygen Saturation- SpO2 using "Photoplethysmography" or PPG.

The World is going through a tough time and  lot of people are infact talking about the scope of home monitoring and remote monitoring.

In this blog post I would like to explain how SpO2 is calculated using a cheap PPG sensor.

The setup :

• BeagleBoneBlack Rev 3 (https://beagleboard.org/black)
• MAX30100 (https://www.maximintegrated.com/en/products/sensors/MAX30100.html)

MAX30100 is a PPG sensor manufactured by Maxim Integrated. The sensor features RED(660nm) and IR(880nm) LEDs , a photo detector  and a processing unit.

Photoplethysmography (PPG) is a simple optical technique used to detect volumetric changes in blood in peripheral circulation. It is a low cost and non-invasive method that makes measurements at the surface of the skin. (https://www.news-medical.net/health/Photoplethysmography-(PPG).aspx )

Photons will be emitted from one or multiple sources to the human tissue. (generally focusing to places where the concentration of arterioles are high. Like Finger tips and ear lobes) Light emitted will either get absorbed by the tissue or reflected back. This photons will be picked up by a set of photo detectors.

There are two main types sensors available

• Reflective
• Transmissive

MAX30100 is a reflective PPG sensor. Which means the light source and the light sensor lies in the same horizontal plane. The light emitted from the source get reflected from the target and monitors at the light sensor.

MAX30100 has an IIO driver available for linux(https://github.com/torvalds/linux/blob/master/drivers/iio/health/max30100.c ). This is the reason why I chose Beaglebone black as the processor board. So I wired up the max30100 to my BBB.

MAX30100 uses I2C bus and the module address is 0x57.

Beaglebone black has 3 I2C buses. Unfortunately I2C 0 (The one which is enabled by default )can't be used because an EEPROM with the same address is present at the board.

So enable the I2C1 and register max30100@57

&i2c1 {
        max30100@57 {
                reg = <0x57>;
                compatible = "maxim,max30100";

                pinctrl-names = "default";
                pinctrl-0 = <&max30100_pins>;

                maxim,led-current-microamp =
                        <24000>,  // max RGB led current in μA
                        <50000>;  // max IR led current in μA

                interrupts-extended = <&gpio1 28 IRQ_TYPE_EDGE_FALLING>;
        };
};

Write a device tree overlay and edit the /boot/uEnv.txt

uboot_overlay_addr4=/lib/firmware/i2c1-max30100.dtbo

After the reboot IIO device will appear. Confirm it by checking /sys/bus/iio/devices

two devices got listed one is the ADC which is a platform device enabled by default and the new max30100.

use lsmod command to confirm.

change the directory to iio:device1 and take a look at the internals

scan_elements will contain the channel enable, channel index and channel type

to enable both RED and IR channels write 1 to the "in_intensity_red_en" and "in_intensity_ir_en"

echo 1 > in_intensity_red_en
echo 1 > in_intensity_ir_en

MAX30100 uses triggered buffer method to access the data. Change the directory to buffers and enable the buffer

echo 1 > enable

Now the data will be streamed as a character device to the node created at /dev/iio:device1

read the data to a file.

cat iio:device1 > file.txt

This file can be processed in python.

MAX30100 provides 2 Bytes (unsigned 16 bit ) values of  RED and IR values. A parsing script has been written to analyze the data and calculate the SpO2

import matplotlib.pyplot as plt 
import numpy as np

import sys 

# Open the file
f = open(sys.argv[1], "rb")

# Global buffers for data
red = []
ir = []

while True:
        # reading 4 bytes
        temp = f.read(4)
        if temp != "" and len(temp) >= 4:
                # ir
                ir.append(temp[1] | (temp[0] << 8)) 
                # red
                red.append(temp[3] | (temp[2] << 8)) 
        else:
                break

# get the mean of the Red and IR
red_dc = np.mean(red[200:350])
ir_dc = np.mean(ir[200:350])

# get the ac content
red_ac = abs(max(red[200:350])-min(red[200:350]))
ir_ac = abs(max(ir[200:350])-min(ir[200:350]))

print("RED DC :{} IR DC :{}".format(red_dc, ir_dc))
print("RED AC :{} IR AC :{}".format(red_ac, ir_ac))

R = (red_ac/red_dc)/(ir_ac/ir_dc)
print("Factor R AC/DC red on AC/DC ir {}".format(R))
print("SpO2 :{}".format(110-25*R))

plt.plot(red[200:350])
plt.plot(ir[200:350])
plt.show()   

The sample plot are given below.

Both RED and IR signals have two parts an AC component and a DC component.

AC component represents the volumetric flow of blood.

DC component represents the reflected photos from the different tissue levels.

Now lets talk about SpO2

SpO2, also known as oxygen saturation, is a measure of the amount of oxygen-carrying hemoglobin in the blood relative to the amount of hemoglobin not carrying oxygen. (https://www.homecaremag.com/understanding-spo2-and-normal-oxygen-levels )

Oxyhemoglobin(HbO2)absorbs visible and infrared(IR) light differently than deoxyhemoglobin(Hb),and appears bright red as opposed to the darker brownHb. Absorption in the arterial blood is represented by an AC signal which is superimposed on a DC signal representing absorptions in other substances like pigmentation in tissue,venous,capillary,bone,and so forth. (https://www.ti.com/lit/an/slaa655/slaa655.pdf )

The SpO2 is the ratio of ratios AC over DC of RED and IR signals.

R = (AC of Red / DC of Red)/ (AC of IR / DC of IR

% SpO2 = 110 – 25 × R

Note : In the above code I have used maxima - minima to calculate the AC component. You can use RMS of the AC component to get slightly better results.