1.0.0
ATTENTION
Device described here is student's experimental project. Do not rely on measurements taken by device like this one.
- NUCLEO F446RE
- LCD Keypad Shield
- Honeywell Pressure Sensor ABPMANN005PG2A3
- KOGE mini pump KPM14A 3V
- KOGE electromechanical valve 3V
- Transistor N-channel BS170
- PNP transistor 2n2905
- four-way splitter, control aquarium valve, plastic tubes
- Solder breadboard, connectors
- Hand cuff
During blood pressure measurement by blood pressure meter cuff placed on patient's arm
must be inflated till absolute closing of artery. Then valve is opened and cuff is slowly
deflated (approximetely 3-5 mmHg/s). When pressure in cuff equilibrates pressure in artery
blood starts flow. This pressure can be considered as systolic pressure. When pressure in
cuff drop below diastolic pressure value oscillation in cuff should disappear (what isn't
entirely true as described later).
In this project 180 mmHg was assumed as pressure closing artery so the cuff was pumped up to 180 mmHg. Then pump was turned off and valve was opened after 1 second delay in order to stabilize pressure sensor output.
Pressure sensor used in project has digital output so all operations on signal like filtering must be performed digitally by microcontroller. All signals showed below were computed by stm32f446re microcontroller and acquired with STM Studio. Sampling frequency of pressure signal from sensor equals 250 Hz.
Systolic pressure was determined as value of first peak detected by sign change of first derivative
of pressure signal. Since raw pressure signal contains noise which is significantly increased in
differentiated signal, lowpass filtration is necessary. First choice was 3rd order IIR lowpass filter
with 20Hz cut-off frequency. This filter was insufficient - abrupt changes caused by valve opening
weren't damped enough as showed on a chart below. As a result that abrupt changes were detected as
a first systolic peak.
Usage of 4th order IIR lowpass filter with 20Hz cut-off frequency eliminated this problem, as shown
on a chart below.
On another chart we can see peak, which was detected by applied method as systolic peak.
Theoretically, when pressure in cuff drop below diastolic pressure value oscillation in cuff should disappear. In fact, pulsatile blood flow still cause oscillations in partially air filled cuff. In order to detect diastolic pressure using filtered pressure signal more sophisticated procedures like computing pulse wave parameters is needed.
Therefore diastolic blood pressure was calculated in simplified way, based on Systolic and MAP pressure calculations. A common method used to estimate the MAP is the following formula [3]:
MAP = DIA + 1/3(SYS – DIA)
Where:
* MAP - mean arterial pressure
* SYS - systolic pressure
* DIA - diastolic pressure
Based on this formula DIA pressure was calculated as follows:
DIA = (3*MAP - SYS)/2
Mean Arterial Pressure was determined based on pressure signal after bandpass filtration. One of the the problems was choice of appropriate bandpass filter. Two filters were tested: FIR filter with lower cut-off frequency = 2 Hz and upper cut-off frequency = 10 Hz and IIR bandpass filter with the same cut-off frequencies. Since FIR filter introduces a delay, the output signal is shifted in time with respect to the input. Group delay of applied filter is 83 so value of pressure signal 83 samples from before detected highest peak was considered as value corresponded to highest peak bandpass filtered signal.
As shown in the tables below values of MAP received by this two bndpass filtered signals in most cases differ from each other. MAP values are similar or even identical only when body and respiratory movement during examination were reduced practically to zero. Filtration by IIR filter was probably more sensitive on patient movements. Besides, results gained from signal filtered by IIR filter were in most cases lower and gave really low diastolic pressure values. Furthermore, signal after filtration by IIR bandpass filter is distorted in comparison to signal after FIR filtration. Signal after FIR bandpass filter is more close to what pulse wave waveform looks like. This effect is caused by unequal group delay, nonlinear phase and fact, that computations were perform on single-precision floating-point numbers[4]. IIR filters are very fast and effective but need for resolution may be critical in some cases, particularly with digital filters, when small but significant numbers combine with large numbers [5]. Efficiency of IIR filters in comparison to FIR filters results from feedback which requires appropriate precision of computations. Therefore FIR filter was chosen. Measurements in table below were performed on one person within one hour. Table contains values of MAP and diastolic pressure computed using signal filtered with FIR bandpass filter and MAP and diastolic pressure computed using pressure signal filtered with IIR bandpass filter as well.
| Calculations based on FIR filtered signal | Calculations based on IIR filtered signal | |||
|---|---|---|---|---|
| SYS | MAP | DIA | MAP | DIA |
| 106 | 78 | 64 | 73 | 57 |
| 107 | 81 | 68 | 75 | 59 |
| 111 | 80 | 64 | 69 | 48 |
| 112 | 79 | 62 | 74 | 55 |
| 112 | 77 | 59 | 69 | 48 |
| 107 | 78 | 63 | 78 | 63 |
| 113 | 85 | 72 | 72 | 52 |
| 104 | 79 | 67 | 79 | 67 |
| 113 | 87 | 74 | 74 | 55 |
| 112 | 84 | 70 | 73 | 54 |
Agata Pawlowska agata.pawlowska07@gmail.com
GNU Public License
- PETRIE, J. C., O'BRIEN, E. T., LITTLER, W. A. and D~ SwErr, M. (1986): 'British Hypertensive Society: Recommendations on blood pressure measurements', Br. Med. J , 293, pp. 611-615
- Leslie Alexander Geddes, M H Voelz, Christopher L. Combs, David S. Reiner, Charles F. Babbs: Characterization of the oscillometric method for measuring indirect blood pressure, DOI:10.1007/BF02367308, Published in Annals of Biomedical Engineering 1982
- https://www.ncbi.nlm.nih.gov/books/NBK538226/
- Steven W. Smith: Digital Signal Processing - A Practical Guide for Engineers and Scientists, 2002
- https://www.rane.com/note157.html
- https://universityhealthnews.com/daily/heart-health/blood-pressure-chart-where-do-your-numbers-fit/