ECG data collection and filtering encountered a bottleneck

w494143467

ECG data collection and filtering encountered a bottleneck [Copy link]

At the beginning, the data collected contained a lot of high-frequency noise, such as AC power, electromyography and some other influences, which made the waveform almost invisible. Later, the high-frequency signal was filtered out through low-pass filtering, and the QRS wave of ECG could be clearly seen. The P wave could also be barely seen, but there were a lot of noises, and the T wave and U wave were almost invisible. Attached below are pictures before and after low-pass filtering.

It can be seen that the difference after filtering is still quite large, but the filtered ECG is still not suitable for direct output and needs further processing. However, I don’t know what algorithm to use at present. I tried the mean filter, but the effect is not good, because many filters will reduce the QRS wave, so the displayed effect is still not good. So I hope someone can help provide some more suitable algorithms to make this wave look more comfortable. Thank you!

maychang

The frequency range of ECG signals is quite low, usually the fundamental frequency is less than 2Hz (120 times per minute), and even the 50th harmonic is only 100Hz. However, ECG signals are relatively weak and easily affected by environmental interference, the most serious of which is power frequency interference. It is recommended to use differential amplification, hardware low-pass filtering, and power frequency trapping at the input stage. Do not directly collect data after broadband amplification of ECG signals.

w494143467

maychang posted on 2020-10-17 13:45 The frequency range of ECG signals is quite low, usually the fundamental frequency is less than 2Hz (120 times per minute), and its harmonic is only 100Hz even at the 50th frequency. But ECG signals are relatively...

OK, I will ask this question to our hardware engineer, as I don't know much about hardware circuits.

志心工作室

You should not use the electrode patch to collect data. The skin is relatively dry, resulting in a large impedance and a lot of noise. It is best to use a high input impedance hardware solution, isolate it, and then wipe the measurement site before measurement to reduce contact impedance. The standard ECG frequency response range for monitoring is 0.5Hz-40Hz, and software filtering generally uses an IIR filter.

w494143467

Zhixin Studio published on 2020-11-17 17:28 This should not be the collection method of sticking electrodes. The skin is relatively dry, resulting in a large impedance and introducing more noise. It is best to use a high input impedance hardware solution first, ...

Because we are making smart wearable devices, some of the requirements you mentioned are not easy to meet

志心工作室

w494143467 Published on 2020-11-17 21:13 Because we are making smart wearable devices, some of the requirements you mentioned are not easy to meet

If you are wearing a wearable device, it is best not to use wired transmission (it is easy to introduce power frequency interference) or isolate the power supply. You should be making a watch or bracelet. There is basically no 50Hz interference when transmitting with Bluetooth. The following is the ECG acquisition module software we made here. It only does a simple low-pass filter.

https://bbs.eeworld.com.cn/thread-1149010-1-1.html