ppg_and_ecg_filters

A collection of subroutines and functions for removing noise and processing ECG and PPG signals I found useful.

Some background, my ECG sensor had a sampling frequency of 250hz and my PPG had one of 75hz

Transfer Functions

Chebyshev Lowpass

• Order 2

Butterworth Lowpass

• Order 5
• Baseline removal (sub 0.5-0.67Hz)

Butterworth Highpass

• Order 2
• HF removal (150Hz)

Butterworth Bandpass

• Order 2
• Reasonable passband is 0.5-150Hz. Here's why

Butterworth Bandstop

• Order 2
• In North America, the electrical grid operates at 60Hz

Other Processing Tools

QRS Peak Detection

• My method of finding R peaks is pretty much returns indicies with ECG measurments of >750uV. Theres not really any sense in applying complicated peak detection algs if its a clean signal

Saturation Correction

• My ECG had an upper limit of 50k uV

Moving Median Filter

• This one is more useful for PPG signals. You could use a bandpass+notch but its not really necessary if you're just trying to compute heart rate or something. The R peak detection works here too, just with different bounds

BPM and HR Variance

• BPM is just the number of R peaks per minute. You could use Q or S peaks too I guess, but R peaks are more prominent thus likely give a more accurate heart rate
• Variance just uses a built in numpy function. This is more useful on time frames where you can approximate as being roughly constant. Variance doesn't tell you anything if a person jumps out of bed and goes for crossfit!