Skip to content
Tutorials Center
My learning

Topics

Lesson 33 of 38 ยท Applications

ECG and Power-Line Denoising

Clean a synthetic ECG with a 50 Hz notch, a baseline-wander high-pass and zero-phase filtering, then detect heartbeats and estimate the heart rate in Python.

A biomedical signal is a perfect test of filter design. An electrocardiogram (ECG) records the electrical activity of the heart as a small voltage (about 1 mV) riding on much larger disturbances. Three disturbances appear in nearly every recording: power-line interference at 50 or 60 Hz, baseline wander from breathing and electrode movement (below about 0.5 Hz), and muscle noise at higher frequencies. This lesson builds a simulated ECG, adds those problems, and removes them with the filters you know.

There is no real patient data here: the ECG is synthetic and the lesson is about signal processing, not diagnosis.

A synthetic ECG

One heartbeat is a sum of five Gaussian bumps called P, Q, R, S and T. Repeating the beat every second gives a signal of 60 beats per minute.

Python Generating a test ECG

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The noisy trace wanders by about a millivolt, as large as the R peak itself. A simple threshold would fail.

The three-filter recipe

ECG cleanup chain

High-pass 0.5 Hz Removes baseline wander
Notch 50 Hz Removes mains hum
Low-pass 40 Hz Trims muscle noise

Use zero-phase filtering so QRS shapes stay symmetric.

1 / 2

Phase matters in the ECG: a nonlinear phase filter distorts the shape of the QRS complex and the ST segment, which clinicians read carefully. That is why offline analysis uses zero-phase filtering (sosfiltfilt, forward then backward). Real-time monitors that cannot look ahead use linear-phase FIRs or carefully designed IIRs.

Python Filtering the ECG

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The RMS error falls from 0.42 mV to about 0.08 mV and the R peak is unchanged. The notch removes the 50 Hz line without touching the nearby QRS energy because its Q is high (narrow). Use 60 Hz in countries with a 60 Hz mains supply.

Detecting the heartbeat

The R wave is the tallest, sharpest part of the beat. A standard detector (inspired by the Pan-Tompkins algorithm) band-passes around 5-15 Hz, differentiates, squares and integrates to emphasize the QRS, then finds peaks with a minimum spacing.

Python QRS detection and heart rate

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Every beat in the record is found (35; the 36th R wave in the list falls just after the end of the 30 s record), the detections are within a few milliseconds of the true R waves, and the heart rate comes out at the true 70.6 beats per minute.

What the adaptive alternative offers

A fixed notch removes interference only at a fixed frequency. If the mains frequency drifts (for example on generators) or if the interference is not sinusoidal, use an adaptive canceller with a reference sensor, or adapt the notch frequency.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Why is zero-phase filtering preferred for offline ECG analysis?

    Forward-backward filtering cancels the phase response so waveform shapes are preserved, which matters for clinical interpretation.
  2. 2. A high-pass filter at about 0.5 Hz is used to remove

    Baseline wander is very slow drift below about 0.5 Hz; a gentle high-pass removes it without touching the heartbeat.
  3. 3. Why is the notch filter given a high Q?

    A narrow notch removes the 50 or 60 Hz line while leaving neighboring ECG frequencies almost untouched.
  4. 4. What is the purpose of squaring in a QRS detector?

    Squaring the derivative makes the signal non-negative and emphasizes the steep QRS slopes relative to smaller waves.

Summary

  • ECGs are corrupted by mains hum, baseline wander and muscle noise; each has its own filter.
  • Use a high-pass near 0.5 Hz, a high-Q notch at 50/60 Hz and a low-pass near 40 Hz, applied with zero-phase filtering offline.
  • A band-pass, derivative, squaring and integration chain highlights QRS complexes for peak detection and heart-rate estimation.
  • Fixed notches fail for drifting interference; adaptive cancellers do not.

Next: AM and FM Demodulation