We use the analogue output of the Portapres device for monitoring continuous blood pressure.
Is band-pass filtering necessary for such blood pressure wave analysis?
What should be the low/high pass cut offs?
Please read this document about your question its figure 3 present a useful comparison of the effects of a low-pass filter in comparison to a high-pass filter.
dear Ali, please indicate the sample frequency of this device, or send me its datasheet.
have a nice day
Dear Ali,
I do not understand why you going to filtrate signal from the Portapres. This device gives usually very clear record. I do not understand your question. Sorry.
Evgeny
Dear Ali,
a suitable cutoff frequencies also depend on the contents you want to assess in you blood pressure curve. If you want to analyse subtle morphological beat-to-beat changes a higher cutoff frequency of the applied low pass filter might be necessery. If you search for "Physics of Invasive Blood Pressure Monitoring" you'll find some hints and references which are directed at the frequency content of blood pressure signals. However, if you intent suchlike analysis you should also pay attention on the functioning of your Portapres device, i.e. which processing is done to derive the waveform which is analogously output.
Sebastian
Dear Evgeny,
The Portapres recorded file on the memory card is quite clear, of course it's at 100 Hz. As we use the analogue output to synchronize the BP signal with other signals (e.g. ECG) we have some noise with the BP signal (a noise of equal to +- 3 mmHg). The Beatscope which is the Portapres software also uses a low pass filter to reduce the noise during signal processing (wavelet analysis). But, as we do not use the Beatscope software first we must filter the signal.
Dear Sebastian,
We are going to apply Modelflow for extracting SBP, DBP, SV, and PVR. Is there a component in BP signal with higher than 30 Hz frequency? If not, so applying a low pass filter of 50-60 Hz shouldn't distort the signal from the actual values. of course a more conservative filter would be 100 Hz.
Hi Mohamad,
The Portapres sampling rate is 200 Hz, of course we recorded all of our signals at the same rate of 1 KHz.The device measures BP continuously based on the volume-clamp method so there are some high frequency non-BP related noise like "50 Hz hum or tape flutter" etc.plus the noise at the analogue output.
The main question is whether there is a high frequency component in BP wave signal which we miss by applying a low pass filter of for example 50 Hz, and if applying such filter distorts the values more than reducing the noise? As we are going to calculate baroreflex sensitivity with sequence method we need a precision of ~ 2 mmHg, this is while the Portapres analogue output resolution is 2.5 mV (mmHg) !
Dear Ali,
i guess keeping content to 30 Hz is fine for many kinds of analyses.
Good luck, Sebastian
Dear Ali
Noise +-3 mmHg is significant. This noise is at the frequency of 30 Hz and higher. I think, you need to filter out this noise to improve the measurement of sequence baroreflex and apply Modelflow.
It looks to me, the problem can be solved by calculating spectra of the BP signals. The spectra should show the border for filtration.
Evgeny
Dear Evgeny,
Yes the noise is significant and we'll try to find the source. Of course if this is due to Portapres analogue output resolution (2.5 mvol) it cannot be reduced by other than filtering I guess.
It's a very good suggestion to look at the BP spectra to find the noise frequency. I'll update you about it.Tnx.
Attached is the PSD analysis of the BP wave signal. If I understood well this shows that frequency of BP related components is almost below 30 Hz. There is a component at ~50 Hz which is very likely the line noise. Applying a low pass filter with 30 Hz (or even 40 Hz) cutoff eliminate the noise while does not distort the values.
You are absolutely right, but you need to check connection of the Portapres output with next device. I think, this noise is because the connection does not have good ground, there is technical problem.
Evgeny
Here the digital filter result designed in MATLAB with the following script:
-----------
%function Hd = lowpassfilter
Fs = 1000; % Sampling Frequency
N = 20; % Order
Fc = 40; % Cutoff Frequency
flag = 'scale'; % Sampling Flag
win = hamming(N+1);
b = fir1(N, Fc/(Fs/2), 'low', win, flag);
Hd = dfilt.dffir(b);
grpdelay(Hd,2048,Fs); % plot group delay
D = mean(grpdelay(Hd)); % filter delay in samples
FilteredBP = filter(Hd, [OriginalBP; zeros(D,1)]); % Append D zeros to the input data
FilteredBP = FilteredBP(D+1:end); % Shift data to compensate for delay
plot(OriginalBP);
hold on;
plot(FilteredBP);
hold on;
-----------
of course we'll also try to reduce the noise by better connections.
Thanks for all your responses
Excellent, Ali. Now you can work with peace in mind.
One note: your picture shows 300 - 400 Hz noise. This could be from a strong generator working close to your lab.
Evgeny
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Biosignals
- Biomedical Signal Processing