topic Re: Analysis of ECG signal in Life Science

Analysis of ECG signal

RODAWG — Wed, 01 Dec 2004 19:21:04 GMT

Can anyone advice me on how to get a display and beats/min from a signal coming out of ECG???
I am receiving 3 signals from the microcontroller (after they are converted by 8 bit A/D) and one of them represents heart rate.
How can I convert that signal to display and give me a digital heartbeat/min value using labview???

Re: Analysis of ECG signal

rpursley8 — Wed, 01 Dec 2004 21:17:42 GMT

I have attached a VI that should do what you want if you have Labview 7.1 and Advanced Analysis Library.

If not, then this is what the VI does.

First you process the signal a little to enhance the R wave.

1. Take the derivative of the signal (difference function), name this signal X
2. Perform a Hilbert Transform of X to get Y
3. Combine to get Z = X + iY (complex) and take the magnitude of Z
4. This process will enhance the R wave amplitude in relation to the rest of the signal

Then you perform a peak detection to find all of the peaks and their locations.

Find the difference between the peaks to get the number of points between beats.

Divide this number by (Sampling Rate * 60) and invert the number to convert to beats/min.

The Advanced Analysis Library has the Derivative, Peak Detect, and Hilbert Transform functions used in the VI. I tried to save it as version 7.0 but it wouldn't let me.

Re: Analysis of ECG signal

RODAWG — Wed, 01 Dec 2004 21:28:41 GMT

Randall,
Thank you so much for your quick response and attached example.
I will try that and update you on what happened.
I assume that the input signal representing the heart beat is a digital signal coming out of the microcontroller in 8 bits via a serial cable as that is the signal I am working with.
Thanks again.
Rodawg

Re: Analysis of ECG signal

RODAWG — Wed, 01 Dec 2004 21:29:13 GMT

Re: Analysis of ECG signal

rpursley8 — Wed, 01 Dec 2004 22:13:26 GMT

What I sent you would as post processing. If you want to use it as real-time processing you would have to change the functionsx to their pt-by-pt versions and it should still work with a little modification. The VI was adapted from some work I did a couple years ago in which I was processing saved data.

Re: Analysis of ECG signal

RODAWG — Thu, 02 Dec 2004 19:39:06 GMT

Thanks again for your prompt response.
Could you advise me what was the format/representation of your data???
Looks like you were storing the data manually in the array you have.
Do you have any samples that I could look and use for this simulation to see and understand the logic and the vi more.
Thanks and have a great day.

Re: Analysis of ECG signal

rpursley8 — Thu, 02 Dec 2004 19:42:49 GMT

Here is a file with 30 seconds of test ECG data (60 bpm).

Re: Analysis of ECG signal

johnsold — Thu, 02 Dec 2004 21:40:12 GMT

Here is another VI which may do some of what you are trying to do. EKG analysis can be very complex - many papers have been published on the subject.

Lynn

Re: Analysis of ECG signal

RODAWG — Fri, 03 Dec 2004 17:41:48 GMT

Thanks alot Randall. Appreciate your help.

Re: Analysis of ECG signal

RODAWG — Fri, 03 Dec 2004 17:42:19 GMT

Thank you Lynn. You have helped me quite alot in my previous problems as well.

Re: Analysis of ECG signal

Luanna — Wed, 07 Jun 2006 01:29:44 GMT

Hello, Randall.
I´ve been using your EKG Processor vi and when i input some of my data i noticed that the EKG peaks appear to have their Time Location shifted, mainly at the end. I used that formula Time[i] = t0 + dt * Location[i] and i plot my waveform using time location instead of just location. Have you experienced that? I´ve attached a part of the waveform.
Please, if you could help me, i´d very much appreciate it.
Regards,

Luanna

Re: Analysis of ECG signal

Luanna — Wed, 07 Jun 2006 01:31:04 GMT

Sorry, forgot to send the image.

Regards,

Luanna

Re: Analysis of ECG signal

rpursley8 — Thu, 08 Jun 2006 20:47:56 GMT

Take a look at where the peak is in the Enhanced EKG signal. Is it also shifted with respect to the input signal?

If it is then the shift is just a result of the processing and as long as it is about the same, your heart rate measurements are fine.

Re: Analysis of ECG signal

Luanna — Tue, 13 Jun 2006 00:01:23 GMT

Randall, i´ll test what you asked (the enhanced signal) but i noticed that on the original ECG signal the peak locations are shifted since the beginning, but it gets worse as the signal advances in time, that is, the shift is not always the same size.

Actually, my main corcern is not this shifted time locations, but the power spectrum. I ran your example and i dont know if im wrong, but shouldnt the spectrum have just one major peak?

Thanks again for your help!
Regards,

Luanna

Re: Analysis of ECG signal

rpursley8 — Tue, 13 Jun 2006 16:41:26 GMT

Can you post an example of your data? I don't see how the spacing between peaks can increase, the enhanced signal is generated from the original.

Re: Analysis of ECG signal

Luanna — Wed, 14 Jun 2006 01:31:08 GMT

Hello, Randall. Here it goes a piece of data. Thank you!

Regards,

Luanna

Re: Analysis of ECG signal

rpursley8 — Thu, 15 Jun 2006 02:25:31 GMT

I think you forgot to attach the data.

Re: Analysis of ECG signal

Luanna — Thu, 15 Jun 2006 16:41:05 GMT

Sorry again...i´ve attached it on the other thread and i forgot this one. Here it goes!

Thank you!

Luanna

Re: Analysis of ECG signal

rpursley8 — Thu, 15 Jun 2006 21:10:40 GMT

Attached is a copy of the EKG Processor vi with your data in it. I may have modified it some to get the heart rate. The results look good to me. I had to modify the level for the peak detection to eliminate a false peak from the glitch in the middle of your data. I don't not see any shift between the original data and the enhanced data.

Re: Analysis of ECG signal

Luanna — Sun, 18 Jun 2006 22:44:26 GMT

Hi, Randall. Thank you for the vi. Considering i´m using almost exactly your algorithm with the difference that instead of Locations i calculate Time Locations, do you think that this formula TimeLocations[i] = t0 + dt*Locations[i] may be the cause of the shift?

Regards,

Luanna