In case you have enough paired samples.

I'm sorry, but could you please explain?

Yes, what you propose would give you a measure of coherence between HRV and respiration. We calculate something similar to this when we assess heart rate/respiration rate coherence. This can be easily done with software suites like WinCPRS.

Know though, that the peak of HRV at 0.1Hz in and of itself is enough to indicate how strong the coherence was, because ultimately when doing HRV BFB, changes in HRV are mostly an effect of the respiration rate. We have recently started measuring the height of the HRV peak at 0.1Hz for each individual study participant, and using that as an indicator of coherence in HRV BFB training. I can't recall if we've published anything yet using this technique, but if you're interested I'll check back through our recent articles for you and send a reference if we have one.

The one caveat to that is because everyone's resonance frequency is slightly different, if participants are doing HRV BFB and breathing at their resonance frequency (i.e., not just breathing with a pacer at 6 breaths per minute), the peak will not necessarily be exactly at 0.1Hz. Because we assess participants' resonance frequency at the start of HRV BFB training  we can measure the height of the HRV spectral peak in the appropriate place. Even this may be superfluous though. Ultimately, the height of the largest peak close to 0.1Hz is an excellent indicator of strength of coherence between HRV and respiration.

Sorry for the delay, I need some time to think over your task. Two questions:

1. Could we suppose that "instantaneous" HR is frequency modulated by respiration and some "noise"?

2.  Is it possible for you to record respiration signal and "instantaneous" HR as time series?

Thanks

Excellent answer by David Eddie. I would only suggest to consider a very good device to collect RR samples as only few beats lost (or extasistolic beats) can deeply influence the spectrum coherence.

1) of course "istantaneous RR  length (hence vatiability) is modulatae by respiration

2) It is possible and already described in literature (se as examples papers by Luciano Benrardi  in the late 90s). It could be pleonastic to remind you that the frequency respiration will coincide with HRV at 01 Hz only when respiratory acts are 6 per min. But in my experience it is not so easy to drive the respiration at this frequency in non well motivated patients,. Hence the cross correlation betwwen RF and HRV spectrums during free respiration is the best way to understand the relationship between the cited physiological acts.

Good luck

fb

It is important to understand for what reason you calculate RRI/respiration coherence.

Coherence is a function of the frequency.  Cross-spectral analysis (transfer function analysis) of RRI/Respiration volume provides an assessment of coherence at all frequencies in the range of HRV spectrum, i.e., gives the value of coherence between HR and respiration at any frequency in this range. WinCPRS software and free “KuBIOS HRV” software allow perform the Cross-spectral analysis in case there are RRI and respiratory curve records.

Hi Dominik, 

There are some things that need to be clarified here. 

1. Coherence means something very specific in this context. In order to actually be measuring coherence, you would need the phase of RR intervals along with its power spectrum. The Fourier transform is how you obtain the power spectrum, and the imaginary part of the Fourier transform is the phase of your signal. 

2. The 0.1 Hz frequency in HRV does not correspond to respiration (at least not natural), it corresponds to Mayer wave induced variability. Mayer waves are caused by natural fluctuations of arterial blood pressure. People naturally breath at around 0.3 Hz. These 2 peaks are easily seen in most RR interval power spectra. Any scientist who is making their subjects breath at 0.1Hz is making a very bad mistake, as the Mayer wave peak and the breathing peak will become undistinguishable.

If you are interested in giving biofeedback on breathing rate and your subjects are breathing super slow, you cannot use RR interval power spectrum. In the case that they are breathing normally around 0.3 Hz, in my experience, you'll be able to identify this frequency in only around 60% of people. You will find plenty of articles saying that the HRV spectrum correlates with breathing, which is true for populations, but not so much for individuals. 

http://en.wikipedia.org/wiki/Coherence_%28signal_processing%29

http://en.wikipedia.org/wiki/Mayer_waves

Hi Dr. Laurin, 

I agree it would be highly problematic to instruct study participants to breathe at 0.1Hz in most physiological or psychophysiological study paradigms, however, in heart rate variability biofeedback this is in fact the goal. Patients are instructed to breath at their resonance frequency (slightly more or less than 0.1Hz) to strategically effect HRV/respiration rate coherence. A growing literature shows regularly practicing resonance frequency breathing has salutary psychosocial effects.

I am using the Kubios Software and recording with the Polar Watch RS800CX. EDR is not enabled in Kubios, probably because I don't have a multi channel ECG...? Is WinCPRS a free software (I didn't find a download link so far) and would I be able to get the EDR with this program, although I am merely recording with the Polar Watch? I hope the question makes sense... 

What is the EDR other than the High Frequency peak (of the autoregressive frequency spectrum) - or the LF peak in case of HRV BFB? 

David, I would be very interested in an article where you use the height of the 0,1Hz peak as an indicator of coherence. What would be the difference between measuring the height of the peak and the LF-Power (ms2)? 

Thank you.

Hi Dr. Fuchs,

If you're using a Polar watch, I'm guessing you are just looking at HRV recordings under normal breathing conditions (i.e., not during HRV BFB). Is this the case? I think a lot of the answers to your questions will depend on what you're studying.

When doing HRV BFB, I would speculate that the height of the peak around 0.1Hz is a clearer indicant of the effects of resonance frequency breathing than the entire area under the LF spectrum. This is because the entire area under the LF spectrum potentially captures a lot of other physiological effects/phenomena.

However, when individuals are not being stimulated at 0.1Hz (such as under normal breathing conditions), the peak at 0.1Hz is probably much less relevant, and total LF power would be a better statistic to look at. I'd be very interested to hear Drs. Vaschillo and Laurin's opinions on this. They may have divergent views. 

Unfortunately WinCPRS is not free software. It comes from Absolute Aliens Oy, who are based in Finland. I believe you'd be looking at about US$5000 for the package. Because we record respiration in our lab, I've never tried to calculate EDR using WinCPRS. It could possibly be done, but I'm not sure. Also, because I've never used EDR, I do not know how it's calculated. Maybe someone else on this thread knows the answer to this. 

I'm attaching an article that introduces the 0.1Hz peak amplitude methodology.

Thank You David Eddie. I understand. But yes, I am doing HRV BFB. The HRV is constantly recorded with Polar Watch. We then extract 5-Minute-Intervals of the baseline, stress-test (iMAST), intervention and postline. During the intervention however, we also record heart rate (BVP) and respiration for the feedback with the program Biotrace, which allows us to extract respiration rate and a measure of coherence (1 to -1). The breathing at around 0,1Hz is facilitated by a Music Therapist via Monochord (Inhalation) and Chanting (Exhalation), which seems to be a easy to understand and well received way of pacing. Though the Music Therapist sees and works with the visual feedback (of Biotrace), I am not sure if it is accurate to call the intervention (single session) HRV Biofeedback, since it is more like a HRV/RSA Training.

How do you measure the height of the 0,1Hz peak? If the answer is in the article, I am sorry, I have yet to read it. 

Very interesting. Can't recall if it explicitly states in the article how one measures the height of the peak. It's a simple manual process though — we simply record by hand the exact amplitude of the peak.

To do this in WinCPRS one just clicks on the top of the peak and it shows the exact amplitude. I'm sure other software packages allow you to do similar. We then simply record the number in our dataset.

What would you say are the easiest and best ways to measure Respiratory Sinus Arrhythmia"? In the article of Grossman and Taylor (2007), I understand that the RSA is the "spectral power of high-frequency RRI power (ms2)". Are the measurements of coherence and RSA essentially the same?

Yes, most agree that HF HRV is a good indicant of RSA, however, as Grossman and Taylor note, HF HRV (which reflects vagal tone) can be influenced by many other factors besides RSA. So to say HF HRV = RSA would not be correct.

Of course, they are very highly correlated, but will be correlated differently under different conditions. For instance, if you were showing study participants aversive picture images, Porges' poly-vagal theory would say that top-down neurological processes may also be affecting HF HRV. Alternatively, during resonance frequency breathing, spectral peak amplitude around 0.1Hz would be a very good proxy for RSA.

Hi, i am here to ask questions rather than answering your question. Hope you do not mind. Do you know how the bio-trace compute the respiration rate, because i cannot find detailed information in the Bio-trace manual. 

As you can see in the attachment, it seems that the respiration rate is not so in accordance with the original respiration data... 

Or i might be wrong..

I computed respiration rate by selecting channels--49. RSP-Rate

Please find the settings below:

Can you tell me my understanding of respiration rate here is correct or not?