offline heart rate extraction from ECG signal

[mbinder]

I have acquired raw ECG signals with EXG electrodes and now I'd like to extract /convert them into HR values.

Do you know any NI VI or any easy-to-implement algorithm for an offline HR data extraction from analog ECG data acquired with BioSemi ActiveTwo?

Any help greatly appreciated!

Marek

[Coen]

A VI to calculate the heartrate from 2 EX channels is available for download from http://www.biosemi.com/download/Heartrate.vi , I have written this VI a few years ago for online heartrate calculation (was included in some custom versions of ActiView), but you can of course use it in off-line applications just as well.

Best regards, Coen (BioSemi)

[Pascal]

Hello,
I had quiet the same question as mbinder. I acquired ECG signals and are now trying to calculate them into HR values. Since I never used Labview before, I have some problems in reading and usign the VI. Is there a "how to" or sth similar I could use.

kind regards
pascal

[Teuniz]

You can also use EDFbrowser to extract heartrate values from raw ECG-data.

http://www.teuniz.net/edfbrowser/

[echoinsist]

Hi Coen,
The HR value you mentioned is heart rate variability? and how to use the VI to extract? Thanks in advance.

Best,
Qian

[liming]

Hi
I am using EX8 electrode and CMS,DRL as reference to get my ECG data. However, i had some problem. The ECG diagram is good and R peaks were obvious in Actiview Two but when i receive the voltage data by c++ algorithm ,it seems not so clear,such as -35768, -25674,- 9.25596e+61，3.4825567e+06. As far as i know, the ECG voltage ranges from -0.6 to 0.6 mv. How could i get the correct data?
This is my c++ code acquiring ECG data from Actiview Two：
（32+8EX channels,256Hz）


buffer_idx = 0;
char data[MSGSIZE];
recv(sock, data, MSGSIZE, 0);

for (int channel = 0; channel < (32 + ; ++channel){
complex<double> sample[40];
double samp[40];
for (int m = 0; m < 2; ++m){
int offset = 3 * (m * (32 + + channel);
samp[channel] = (data[offset + 2] << 16);
samp[channel] += (data[offset + 1] << ;
samp[channel] += data[offset];
sample[channel] = samp[channel];
A[channel].push_back(sample[channel]);

Look forward to your reply.

[Coen]

ECG/EEG/EMG is measured BETWEEN two (or more) electrodes. The signal of a single electrodes with respect to CMS (raw, unreferenced) is not a valid signal, see

https://www.biosemi.com/faq/cms&drl.htm
viewtopic.php?f=1&t=33&p=86
viewtopic.php?f=1&t=40&p=109
viewtopic.php?f=1&t=177

So, for ECG use (at least) two electrodes on the chest and subtratc the signals in sofwtare.

The scaling for active electrode systems is:

32-bit words from USB-reciever: divide by 8192 (32*256) to convert to microVolt
24-bit words in the BDF file: divide by 32 to convert to microVolt

Note: custom systems for passive electrodes have 1/4th of the above gain (divide by respectively 32768 and 128)

Note that the system uses a DC amplifier. Therefore, the ECG/EEG/EMG signal is superimposed on a relatively large (typically several tens of milliVolt) electrode offset voltage. Use a high-pass filter or baseline correction to cancel the DC offset and extract the biopotentials.

Best regards, Coen (BioSemi)

[liming]

Hi, Coen
Thank you for your reply! Now I am using EX7 and EX8 which are placed under my left and right collarbone to acquire ECG data. Reference electrode CMS on my left thumb and DRL on my right thumb, is that connection correct? I used c++ algorithm to get 24bits ECG data from Actiview via TCP/IP. So that is to say, i have to record EX7 and EX8 and calculate EX7-EX8,then the EX7-EX8 divided by（256*24）is final ECG（uv）？And what are the active electrodes and passive electrodes ？

Best regards，Lee

[Coen]

CMS on the center of the chest may give better CMRR, DRL can be anywhere on the body.

Divide the 24-bit results by 32 to scale the values to microVolt.

You are using active electrodes (preamp integrated with electrode), ignore the remark about passive electrodes.

Best regards, Coen (BioSemi)

[liming]

Hi，Coen
Your reply is very useful to me.One last question. In the Bipolar Leads display mode, i could see very good ECG diagram like obvious R peaks and stable T wave(row EX7 - EX8).However, in Monopolor Display mode,the diagrams of EX7 and EX8 look like noise（no obvious peaks and troughs）.Nevertheless， i can only record data of mono-channel. Should i just record EX7 and EX8 respectively without considering their diagrams and use EX7 subtracts EX8 ? Do you have other online method acquire EX7-EX8 in Bipolar Leads display mode directly?
Sincerely Yours

[Coen]

You should indeed record both E7 and E8 (use the "add 8 EX electrodes" in the TCP/IP tabpage) . The noise that you see in the raw E7 and E8 signals is mainly Common Mode (same noise for both channels). When you calculate E7-E8, then the Common Mode noise is canceled out, and the ECG remains (this is what ActiView does for the bipolar leads).

Best regards, Coen (BioSemi)

[Cedric]

Hi Coen,

I had read somewhere that it was recommended to do a downsampling at 64 Hz, a lowpass filter at 15 Hz to remove the noise, and a Highpass filter at 0.1 Hz to remove the drift. Would you agree with these values to preprocess Biosemi ECG data?

Thank you for providing the information above about dividing by 32 to get the accurate microvolt values.

Do you please know where I could find an official reference for one or all of these guidelines to include in a paper I am currently writing?

Kind regards,
Cedric Cannard

[Coen]

I would not agree with the above values because not much ECG information would be left in the outcome. Google on "ECG bandwidth" and you will find bandwidth recommendations in the order of 0.05-100 Hz (so at least 512 Hz sample rate).

The LSB value used in the ActivTwo is publicized on https://www.biosemi.com/activetwo_full_specs.htm

Best regards, Coen (BioSemi)

[Cedric]

Noted. Thanks so much!