Oscillometric blood pressure measurement

Automatic blood pressure monitors use the small pressure changes while deflating the cuff to determine both the systolic and diastolic blood pressure. In this clip we emulate an automatic blood pressure monitor with a standard manually operated cuff. To be able to record the pressure and the small changes we have added a pressure sensor which is inside of the automatic blood pressure monitor and plotted its readings together with the ECG. We checked also against the Omron blood pressure monitor which reported before filming at 10am 92/61,91/59 and after filming at 6pm 103/72,95/66. Plots: 1st trace is Einthoven II, 2nd trace is the absolute pressure, lowpass filtered (5Hz), and the 3rd trace is pressure, highpass (0.5Hz) and lowpass (5Hz) filtered, amplified by x50.

The blood pressure is determined from amplitudes of the oscillations. Remember that this measurement is made by deflating slowly the cuff which means that we have different values for the pressure over time in the first place so that a search algorithm needs to go back/forth in time to determine the blood pressure. Because it needs to go back in time the pressure values need to be stored in the memory of the blood pressure monitor and then can be analysed to obtain the blood pressure. This is now described in detail: First, the maximum amplitude of the oscillations is determined (Δp_max) which corresponds to the mean arterial pressure (MAP). Then from the MAP point a search algorithm moves to higher pressures by going back in time and determines the point where the oscillations have decayed to an amplitude of Δp_systolic/Δp_max=0.55 which represents the systolic pressure. Then the search algorithm moves from the MAP point forward in time until the oscillations have decayed to an amplitude of Δp_diastolic/Δp_max=0.85 which represents the diastolic pressure. However, note that the factors 0.55 and 0.85 have been determined heuristically and there is a wide uncertainty in these values. The ranges are:

Δp_systolic/Δp_max Δp_diastolic/Δp_max
0.45 - 0.57 0.69 - 0.89

The paper (Ursino et al, 1996) below provides an analytical solution which confirms the heuristic values. However the uncertainty is still similar to the heuristic values.

The second source of errors is the measurement process itself. Remember that the pressure changes are very small and that artefacts can easily be generated. After we had recorded the data for this figure we measured the blood pressure with the Omron automatic blood pressure monitor and got: 89/66, 88/62, 108/73 and 100/65. Note the large variation of the readings. This comes now at no surprise knowing that the changes in blood pressure are very small and that every movement of the arm and/or muscle contraction will also add small pressure changes. Because every oscillation is a one-off event it is not possible to distinguish between artefact and blood pressure changes. The only solution is to keep still so that these artefacts are not generated in the first place.

Overall this method can only give reliable results if one averages over a couple of measurements. Most automatic blood pressure monitors offer this feature. However, remember that the algorithm which determines both diastolic and systolic blood pressures is based on heuristic values which introuduce a large uncertainty in the measurement process. Different blood pressure monitors might give different readings because the developers have used different values in their algorithm.

Much more precise (but requires practise) is the traditional blood pressure measurement which is covered in the clip about heart-sounds.

Circuit diagram

The Omron 25MPP-02 requires a constant current of 100uA which we establish by measuring the voltage drop over a resistor. The reference voltage is obtained from a green LED which gives us 2V and, thus, our current sensing resistor needs to be 20K.


We thank Peter Macfarlane for the feedback on the script.

next: Heart-sounds.