A New Method to Determine Systolic Blood Pressure Indirectly Aided

by Parallel Recording of ECG and PPG

Péter Nagy

and Ákos Jobbágy

Department of Measurement and Information Systems, Faculty of Electrical Engineering and Informatics,

University of Technology and Economics, Magyar tudósok körútja 2, Budapest, Hungary

Keywords: Blood Pressure Measurement, ECG, Photoplethysmography.

Abstract: Raised blood pressure severely increases the risk of lethal cardiovascular diseases. Home monitoring of blood

pressure is vital in early detection and treatment of hypertonia. Accuracy of indirect blood pressure

measurement methods is sensitive to many physiological factors that are difficult to measure or control. The

accuracy can be improved by using further sensors. In this paper, we propose a new method for the estimation

of systolic blood pressure based on cuff pressure, ECG and photoplethysmographic (PPG) signals. PPG is

measured without hardware filtering keeping the DC-component and avoiding the problem of distorting the

signal. The proposed method was validated by applying it to healthy senior and healthy young adults at rest

and by making a measurement series containing mild physical exercise for healthy young adults. Results of

the tests clearly show the supremacy of the new method to conventional oscillometric procedure.

1 INTRODUCTION

Hypertension is a major cardiovascular risk factor and

the leading cause of death and disability-adjusted life-

years worldwide (Carey & Whelton, 2018). Early

detection and appropriate management of

hypertension is based on blood pressure (BP)

measurement, which is conventionally carried out by

a healthcare professional in the office. However, out-

of-office BP measurement has been shown to provide

more reliable estimation of the average BP of an

individual over time (Stergiou & Bliziotis, 2011).

Home BP monitoring allows the detection of white-

coat hypertension and masked hypertension and

contrary to ambulatory BP monitoring; it does not

require expensive instrumentation. For home BP

monitoring, automated oscillometric devices are

currently recommended, because they are non-

invasive, easy to use and cheap devices, requiring

little training compared to auscultatory devices

(Stergiou et al., 2018). Despite its widespread use and

many advantages, the oscillometric method has also

major limitations. First of all, the classic method

basically measures the mean arterial pressure (MAP),

systolic blood pressure (SBP) and diastolic blood

https://orcid.org/0000-0003-1902-0469

https://orcid.org/0000-0001-6569-6000

pressure (DBP) values are only calculated

(Drzewiecki, Hood & Apple, 1994). The accuracy of

the calculated SBP and DBP values is sensitive to

several physiological factors including pulse

pressure, anatomical position, elasticity and size of

the measured artery and properties of the surrounding

tissue (Tholl, Forstner & Anlauf, 2004). Increased

arterial stiffness generally impairs the accuracy of

oscillometric blood pressure measurement (van

Popele et al., 2000). Since the introduction of the

oscillometric method, several improved algorithms

have been published including amplitude- and slope-

based methods, model-based algorithms, methods

based on neural networks and machine learning and

the exploitation of extra sensor signals enabling the

calculation of the pulse wave transit time (PWTT).

Forouzanfar et al. presented a review article about the

BP measurement based on oscillometric algorithms

(Forouzanfar et al., 2015). Alghamdi et al. gave an

overview of BP measurement methods based on

machine learning and proposed a classification-based

BP estimation method (Alghamdi et al., 2020).

Despite the major progress in this field, oscillometric

BP measurement techniques still have limitations as

highlighted in (Alghamdi et al., 2020; Forouzanfar et

Nagy, P. and Jobbágy, Á.

A New Method to Determine Systolic Blood Pressure Indirectly Aided by Parallel Recording of ECG and PPG.

DOI: 10.5220/0010891400003123

In Proceedings of the 15th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2022) - Volume 4: BIOSIGNALS, pages 221-227

ISBN: 978-989-758-552-4; ISSN: 2184-4305

221

al., 2015). In the present paper, we focus on the

improvement of the accuracy of cuff-based indirect

SBP estimation. We propose an algorithm for the

detection of the time instant when cuff pressure (CP)

is equal to the maximum arterial pressure (supposed

to be equal to SBP) both during inflation and during

deflation of the cuff using photoplethysmography

(PPG) and the ECG signal. Results of the application

of the algorithm for recordings of healthy senior

adults and healthy young adults are also reported. The

paper does not deal with the application of PPG in

cuffless estimation of blood pressure, where the

research aim is different.

2 MATERIALS AND METHODS

2.1 SBP Estimation Aided by the PPG

Signal

PPG is an optical technique for the measurement of

cardiac-induced pulsatile changes in tissue blood

volume (Elgendi, 2012). Monitoring PPG during BP

measurement helps estimate SBP. If the PPG sensor

is placed on a finger of the same arm as the cuff is

wrapped, the occlusion of the brachial artery by the

cuff influences the PPG waveform. When CP exceeds

SBP, the brachial artery is completely occluded

during the whole heart cycle, and the pulsation in the

PPG signal disappears. During deflation, the

pulsation in the PPG signal reappears when CP falls

below SBP. Accurate designation of the time instants

corresponding to the disappearance and reappearance

of PPG pulsation is not straightforward, especially if

the signal-to-noise ratio is unfavorable. In some

studies, the authors designated these disappearance

and reappearance instants by visual inspection

(Jönsson, Laurent, Skau & Lindberg, 2005; Nitzan et

al., 2005; Nitzan et al., 2013). Visual inspection may

be appropriate for research purposes, but for devices

used in healthcare, automated methods are required.

Lubin et al. considered PPG pulses absent if the value

of the AC signal amplitude was lower than 20 % of

its maximum value (Lubin, Vray & Bonnet, 2020).

Nitzan et al. investigated both the disappearance and

reappearance of PPG pulsation. They considered PPG

pulses to disappear if the value of the maximal

derivative was lower than 1% of the mean initial

maximal derivative. For the detection of the

reappearance of the pulsation, the PPG curve was

divided into time segments corresponding to heart

cycles. Two parameters were calculated in each

segment: an area parameter, which is related to the

pulse waveform and the cross-correlation of the

signal in each segment with the signal in the

neighboring segments. The authors considered PPG

pulses to reappear if the calculated values of the area

and the cross-correlation parameters met certain

detection criteria. The authors measured PPG on the

two index fingers and used both signals for the

calculation of parameters (Nitzan, Patron, Glik &

Weiss, 2009). We have found that the 1% threshold

value of the maximal derivative could be applied to

the recordings only after high-order filtering of the

PPG signal. However, high-order filtering distorts the

signal, which is a potential source of error especially

when the derivative of the signal is small. Therefore,

we did not utilize the derivative of the PPG signal to

detect the disappearance of pulsation. Our proposed

method is new in incorporating both the amplitude

information in AC signal and the correlation between

subsequent heart cycles of the PPG curve, and in

utilizing the DC component of the PPG signal. The

ECG signal recorded in parallel with PPG gives

valuable information when the PPG amplitude is

small.

2.2 New SBP Estimation Method Aided

by the DC-coupled PPG Signal and

the ECG Signal

Our proposed method designates the disappearance

and reappearance of the pulsation in the PPG signal

by applying similar operations independently both

during inflation and deflation. The method is based

on parallel recording of three signals, the CP of the

cuff wrapped around the upper left arm, the PPG

measured on the left index fingertip and the ECG in

Einthoven II-lead. The algorithm consists of three

stages.

At the beginning of the first stage, the PPG signal

is bandpass filtered between 0.5-8 Hz and inverted so

that upward signal corresponds to higher blood

volume in the fingertip. After that, the difference

between the maximum and minimum points of the

PPG signal (AC amplitude) is calculated in a sliding

window. The window length was empirically

determined and set to 500 ms. This window length is

long enough to include the whole systolic upstroke

segment of the PPG curve even in case of low heart

rate. Next, the algorithm locates the intervals both for

inflation (Intinfl) and for deflation (Intdefl) where the

sliding window AC amplitude is permanently below

10 percent of its maximum value.

In the final step of the first stage of the algorithm,

the point is searched for where the sliding window

AC amplitude falls below a threshold level. The

threshold levels are calculated as fractions of the

BIOSIGNALS 2022 - 15th International Conference on Bio-inspired Systems and Signal Processing

222

average AC amplitude over Intinfl and Intdefl. As a

result, a short interval is allotted both for inflation and

for deflation, where the CP equals the maximum BP.

In the second stage of the algorithm, the first local

maximum and local minimum are searched for in the

bandpass filtered PPG signal that are at the beginning

of the search intervals for inflation and deflation.

These points are designated by t

PPGmax_infl

, t

PPGmin_infl

(and t

PPGmax_defl

, t

PPGmin_defl

for deflation). Two more

local maxima are localized preceding t

PPGmax_infl

(in

case of deflation following t

PPGmax_defl

), in order to

delineate two adjacent heart cycles that are required

for correlation calculation. Heart cycles are

delineated based on local maxima; three successive

local maxima enable the delineation of two heart

cycles. For the localization of PPG local maxima and

minima, the ECG signal is also used, as minima and

maxima are searched for within the time interval

between corresponding neighboring R-peaks. After

the identification of heart cycles, the correlation

coefficient (CC) is calculated between the two

adjacent heart cycles in the PPG signal, similarly to

(Nitzan et al., 2009). The shorter heart cycle is

stretched by interpolation in order to have two heart

cycles of equal length. Besides the correlation

coefficient, PWTT is calculated between the

corresponding R-peak in the ECG signal and

PPGmin_infl

PPGmin_defl

for deflation). If CC is larger

than 0.85 and PWTT is between 100-500 ms, then a

valid PPG pulse belongs to t

PPGmax_infl

and t

PPGmin_infl

(or t

PPGmax_defl

and t

PPGmin_defl

in case of deflation).

Otherwise the supposed two adjacent heart cycles do

not exist, the pattern is considered to be noise. In case

of a valid PPG pulse, t

PPGmax_infl

and t

PPGmin_infl

are

moved to the corresponding points in the following

heart cycle; CC and PWTT are also recalculated. In

case of noise, t

PPGmax_infl

and t

PPGmin_infl

are moved to

the corresponding points in the preceding heart cycle;

CC and PWTT are also recalculated. The procedure

continues until the first heart cycle when change in

PPG is considered noise in the direction of following

heart cycles, or valid pulse in the direction of

preceding heart cycles. If one heart cycle is

considered as noise, the algorithm examines the

following two heart cycles and the preceding heart

cycle, calculates CC between all possible pairings of

heart cycles and takes the maximum of these CC

values. This step is necessary to identify irregular or

noisy heart cycles (at most two consecutive heart

cycles) that are followed by at least one valid pulse.

The outputs of the second stage t

stage2_infl_max

and

stage2_infl_min

are the time instants when the PPG signal

has local maximum and local minimum during the

last heart cycle that results in pulsation. In case of

deflation, the direction of moving t

PPGmax_defl

and

PPGmin_defl

is inverted compared to inflation, the first

heart cycle resulting in pulsation is searched for and

the outputs of the second stage are called t

stage2_defl_max

and t

stage2_defl_min

Figure 1: Typical DC-coupled PPG signal (solid line) and

the corresponding CP signal (dotted line). Left vertical line:

estimated time instant where CP exceeds SBP during

inflation. Right vertical line: estimated time instant where

CP falls below SBP during deflation.

In the third stage of the algorithm, DC level of the

PPG signal is also considered. Instead of the bandpass

filtered PPG signal, the lowpass filtered signal is

used; the cutoff-frequency of the filtering is 8 Hz. A

typical DC-coupled PPG signal and the CP signal

recorded in parallel are plotted in Fig. 1.

There is an upward trend in the DC level during

inflation and then the DC level starts falling when CP

is close to SBP during inflation (left vertical line).

The falling trend ends, and the DC level starts rising

again when CP is close to SBP during deflation (right

vertical line). The algorithm detects the points in the

DC-coupled PPG signal, where the trend changes as

possible candidates of time instants when CP equals

SBP during inflation and deflation. These points are

denoted by t

PPG_DC_SBP_infl

and t

PPG_DC_SBP_defl

. Then,

stage2_infl_max

is compared to t

PPG_DC_SBP_infl

and

stage2_defl_min

is compared to t

PPG_DC_SBP_defl

, because the

trend change corresponds to a local maximum during

inflation and a local minimum during deflation. If

stage2_infl_max

is located following t

PPG_DC_SBP_infl

, then

PPG_DC_SBP_infl

is rejected because a valid PPG pulse

was found following it. In that case, t

stage2_infl_max

and

stage2_infl_min

are considered and the time instant of the

next local minimum in the bandpass filtered PPG

signal is extrapolated. This time instant is denoted by

xtrp_infl

. The extrapolation is based on the length of the

previous heart cycle. The time instant where CP

0 10203040

Time (s

)

100

150

200

250

300

PPG DC value (a.u.)

100

120

140

160

180

Cuff Pressure (mmHg)

PPG_DC_SBP_defl

PPG_DC_SBP_infl

A New Method to Determine Systolic Blood Pressure Indirectly Aided by Parallel Recording of ECG and PPG

223

exceeds SBP is estimated as the midpoint between

stage2_infl_max

and t

xtrp_infl

. If t

stage2_infl_max

is located

preceding t

PPG_DC_SBP_infl

, then a distinction is made

depending on whether the distance of the two points

is larger than the corresponding heart period (the

length of the heart cycle) or not. If the distance is

larger, then t

PPG_DC_SBP_infl

is rejected because it is

more than one heart cycle away from the last found

valid PPG local maximum. In that case, t

stage2_infl_max

and t

stage2_infl_min

are considered, t

xtrp_infl

is designated

and the time instant where CP exceeds SBP is

estimated as the midpoint between t

stage2_infl_max

and

xtrp_infl

. If the distance between t

stage2_infl_max

and

PPG_DC_SBP_infl

is smaller than one heart period, then

PPG_DC_SBP_infl

is accepted as the estimate of the time

instant where CP exceeds SBP. In case of deflation,

if t

stage2_defl_min

is located preceding t

PPG_DC_SBP_defl

, or

the distance between the two time instants is larger

than one heart period, then extrapolation is done

similarly to inflation, but in the opposite direction.

The time instant where CP falls below SBP is

estimated as the midpoint between the extrapolated

local maximum in the bandpass filtered PPG signal

and t

stage2_defl_min

. If t

stage2_defl_min

is located following

PPG_DC_SBP_defl

, and the distance between the two time

instants is smaller than one heart period, then

PPG_DC_SBP_defl

is accepted as the estimate of the time

instant where CP falls below SBP. Characteristic

points designated by the algorithm in the bandpass

filtered PPG signal are illustrated in Fig. 2.

Our proposed method aims to estimate SBP with

a temporal resolution better than one heart cycle. This

means that not only heart cycles in the PPG signal are

categorized as valid PPG pulse or noise, but also the

moment when CP equals SBP is searched for within

one heart cycle. If the temporal resolution of SBP

detection is only one heart cycle, the error of the

estimated SBP difference between inflation and

deflation can be high, as demonstrated in Fig. 3.

In the given example, if it is not known, where the

time instant corresponding to SBP is located within

one heart cycle, the estimated SBP difference

between inflation and deflation can be between 5-15

mmHg.

2.3 Data Acquisition

The proposed method was tested on recordings taken

by a home health monitoring device (Nagy &

Jobbágy, 2018). The device contains an inflatable

cuff with two control valves, and it is able to keep cuff

pressure at a constant value. The device also measures

ECG in Einthoven II-lead and PPG at the fingertip

with a sampling frequency of 1 kHz. A transmission-

type PPG sensor is used to reduce the effect of motion

artifacts. The device contains no hardware filters for

ECG and PPG, in order to record undistorted signals.

As a result, DC-coupled PPG signal can be measured.

The device inflates the cuff with approximately 6

mmHg/s speed to 170 mmHg, then deflates the cuff

with the same speed. Deflation is stopped at 60

mmHg for 10 seconds. When 40 mmHg is reached

during deflation, CP changes abruptly to 0 mmHg.

(The CP=constant section makes possible the analysis

and compensation of breathing. Our results in this

field are not reported in this paper.)

Figure 2: Illustration of characteristic points designated by

the algorithm in the bandpass filtered PPG signal.

Figure 3: CP signal during inflation and deflation close to

SBP. If the temporal resolution of SBP detection is one

heart cycle, the error of the estimated SBP difference

between inflation and deflation can be high.

2.4 Tested Persons

5 healthy senior persons (age 55-64 years, 2 females,

3 males) and 13 healthy young persons (age 19-35

years, 6 females, 7 males) participated in the

measurement series reported in this paper. All tested

persons gave their informed consent. The research

PPG AC value (a.u.)

Cuff Pressure (mmHg)

BIOSIGNALS 2022 - 15th International Conference on Bio-inspired Systems and Signal Processing

224

was performed in accordance with the Declaration of

Helsinki and the study protocol was approved by the

Scientific and Research Committee of the Hungarian

Medical Research Council (SE RKEB 46/2020).

2.5 Measurement Protocol

In order to validate the proposed new algorithm, three

measurement series were used.

SBP values calculated by the algorithm are

denoted by SBP

infl

during inflation and SBP

defl

during

deflation. We denote the difference between SBP

infl

and SBP

defl

by ΔSBP as defined in (1).

ΔSBP = SBP

infl

–

SBP

defl

)

The first measurement series assessed the effect

of physical stress on ΔSBP. Each person performed

20 squat jumps. 8 healthy young adults participated

in the measurement series. For each person, one

measurement was recorded by the home health

monitoring device before the exercise and

immediately after the exercise, in sitting position.

The second test series compared the ΔSBP

between seniors and young adults, measurements of

5 healthy seniors and 5 healthy young adults were

analyzed. All measurements were recorded in resting

state of the tested person, in sitting position. 3

measurements were recorded for each person.

The third measurement series aimed at testing the

within-subject variability of ΔSBP, measurements of

4 healthy young adults were analyzed. All

measurements were recorded in resting state of the

tested person, in sitting position. 8 measurements

were recorded for each person.

3 RESULTS

Table 1 – 3 show results for the effect of physical

stress, ΔSBP difference between seniors and young

adults and within-subject variability, respectively.

The proposed algorithm gives a good estimate of

SBP

infl

and SBP

defl

. The reference SBP

infl

and SBP

defl

values (used as gold standards) were determined by

visual evaluation: the CP value when pulsation in

PPG disappears or reappears. Based on the ECG R

peak position we could restrict the range within every

heartbeat where pulsation in PPG can be present. This

helps especially when the signal-to-noise ratio is low.

50 recordings were selected for visual analysis to

evaluate the accuracy of the algorithm including 5

recordings after physical exercise, 15 recordings from

healthy seniors and 30 recordings from healthy young

subjects.

Table 1: The effect of physical stress.

Person

Identifier

ΔSBP

(

mmH

Before

exercise

After

exercise

Change

Y1 4.4 -2.8 -7.2

Y2 7.1 5.6 -1.5

Y3 -4.2 9.3 13.5

Y4 3.7 -1.3 -5.0

Y5 11.9 12.7 0.8

Y6 1.7 -0.8 -2.5

Y7 7.1 4.8 -2.3

Y8 -3.8 4.3 8.1

Y1 4.4 -2.8 -7.2

Y2 7.1 5.6 -1.5

a. Eight healthy young subjects.

Table 2: Senior-young group average.

Group

ΔSBP

(

mmH

Mean

Min Max

Senio

-1.2 -11.5 10.6

Young 4.1 -5.0 11.6

a. Five senior healthy subjects, five young healthy

subjects.

b. Three measurements taken for each subject, sitting

position, resting state.

Table 3: Within-subject variability.

Person

Identifier

ΔSBP

(

mmH

Mean

Min Max

Y9 5.1 -0.1 8.2

Y10 5.8 -5.0 14.2

Y11 8.3 -0.1 13.4

Y12 7.0 -3.3 15.2

a. Four young healthy subjects.

b. Eight measurements taken for each subject, sitting

position, resting state.

4 DISCUSSION

The values determined by the suggested new

algorithm are within a narrow range around the gold

standard. The mean value and standard deviation of

the difference between values determined by the

algorithm and the gold standard was -0.7 ± 1.5 mmHg

for SBP

infl

and -0.3 ± 1.8 mmHg for SBP

defl

. In 88 %

of the 50 examined recordings, the absolute error of

the algorithm compared to the gold standard is below

3 mmHg. Considering the 6 mmHg/s speed of

inflation and deflation, this corresponds to a temporal

resolution much better than one heart cycle. The

proposed new algorithm assures better than usual

resolution in SBP determination. This is necessary for

the

appropriate evaluation of the three measurement

A New Method to Determine Systolic Blood Pressure Indirectly Aided by Parallel Recording of ECG and PPG

225

series.

CP values corresponding to time instants when CP

is supposed to be equal to SBP indicated by the PPG

signal may considerably differ during inflation and

deflation. Both for seniors and for young adults, the

largest absolute difference found was more than 10

mmHg. This difference cannot be evaluated using the

usual indirect BP measurement methods. The method

proposed in this paper assures the necessary

resolution and accuracy.

The effect of physical stress on ΔSBP showed

large variability among individuals, in both

magnitude and direction. The change in ΔSBP was

negative for 5 persons and positive for 3 persons, with

the smallest absolute value of 0.8 mmHg and the

largest absolute value of 13.5 mmHg. Sign of ΔSBP

changed for five persons because of exercise.

Interestingly, the person showing the largest ΔSBP

value before exercise (Y5) showed the smallest

change in ΔSBP because of exercise. ΔSBP gives

information basically on the circulation in the upper

arm.

Group-level comparison of senior and young

adults revealed negative mean ΔSBP value for seniors

and positive mean ΔSBP value for young adults.

Although negative ΔSBP values appeared also for

young individuals, this result suggests that people

with increased arterial stiffness tend to produce more

negative ΔSBP values.

ΔSBP values showed high within-subject

variability (more than 8 mmHg difference of the

minimum and maximum values) for all four tested

persons although the mean ΔSBP was higher than 5

mmHg for each of them. This result shows that one

single measurement of ΔSBP may not be sufficient to

estimate the state of the cardiovascular system of a

tested person.

DC level of the PPG signal contains valuable

information aiding SBP estimation. However, valid

pulses were found following t

PPG_DC_SBP_infl

and

preceding t

PPG_DC_SBP_defl

in some cases suggesting

that the DC level of PPG is influenced by more

factors. Occlusion of the artery by the cuff is a

dominant factor as DC level shows a rising trend

during inflation in all cases.

We compared our results with the work of Zheng

et al. who investigated the mechanical behavior of the

brachial artery during cuff inflation and deflation

(Zheng, Pan & Murray, 2013). The authors compared

SBP, DBP and MAP values measured during

inflation and deflation of the cuff by manual

auscultatory and automated oscillometric methods.

SBP from inflation was found to be statistically

significantly lower than SBP from deflation,

measured by both the manual and the automated

method. This result was not justified by our

measurements. The difference in SBP during inflation

and deflation was studied also by other researchers

(Fabian et al., 2016).

SBP

infl

and SBP

defl

are close to each other.

Comparing them needs a measurement method with

good resolution. In this regard, the method we

propose outperforms other methods. It finds the

disappearance and reappearance of the PPG signal

with better time resolution than a heartbeat while

other methods can detect these fiducial points with

heartbeat resolution only.

5 CONCLUSIONS

Accurate indirect BP measurement is the cornerstone

of the detection and management of hypertension. In

this paper, we proposed a new method for the

estimation of SBP using CP, ECG and DC-coupled

PPG signals. It eliminates the most severe device

dependent limitation associated with indirect BP

measurement, thus our method offers more accurate

non-invasive estimation than the presently available

other methods even in the case of arrhythmia. In the

near future, the algorithm will be applied to patients

with Left Ventricular Assist Device, LVAD. The non-

invasive blood pressure measurement of these

patients is important but cumbersome, requiring

special expertise. Our proposed method is applicable

for them also at home, without medical expert.

ACKNOWLEDGEMENTS

The research reported in this paper and carried out at

the Budapest University of Technology and

Economics was supported by the “TKP2020,

National Challenges Program” of the National

Research Development and Innovation Office (BME

NC TKP2020).

REFERENCES

Alghamdi, A. S., et al. (2020). A novel blood pressure

estimation method based on the classification of

oscillometric waveforms using machine-learning

methods. Applied Acoustics, 164(5), 107279.

Carey, R. M. & Whelton, P. K. (2018). Prevention,

detection, evaluation, and management of high blood

pressure in adults: synopsis of the 2017 American

College of Cardiology/American Heart Association

BIOSIGNALS 2022 - 15th International Conference on Bio-inspired Systems and Signal Processing

226

Hypertension Guideline. Annals of Internal Medicine,

168(5), 351-358.

Drzewiecki, G., Hood, R. & Apple, H. (1994). Theory of

the oscillometric maximum and the systolic and

diastolic detection ratios. Annals of Biomedical

Engineering, 22(1), 88-95.

Elgendi, M. (2012). On the analysis of fingertip

photoplethysmogram signals. Current Cardiology

Reviews, 8(1), 14-25.

Fabian, V., Havlik, J., Dvorak, J., Kremen, V., Sajgalik, P.,

Bellamy, V., Schirger J.A., Sovka P. & Johnson, B. D.

(2016). Differences in mean arterial pressure of young

and elderly people measured by oscilometry during

inflation and deflation of the arm cuff. Biomedical

Engineering/Biomedizinische Technik, 61(6), 611-621.

Forouzanfar, M., et al. (2015). Oscillometric blood pressure

estimation: past, present, and future. IEEE Reviews in

Biomedical Engineering, 8, 44-63.

Jönsson, B., Laurent, C., Skau, T. & Lindberg, L. G. (2005).

A new probe for ankle systolic pressure measurement

using photoplethysmography (PPG). Annals of

Biomedical Engineering, 33(2), 232-239.

Lubin, M., Vray, D. & Bonnet, S. (2020). Blood pressure

measurement by coupling an external pressure and

photo-plethysmographic signals. In 2020 42nd Annual

International Conference of the IEEE Engineering in

Medicine and Biology Society (EMBC). IEEE.

Nagy, P., & Jobbágy, Á. (2018). Personalization of the

oscillometric blood-pressure measurement. In World

Congress on Medical Physics and Biomedical

Engineering 2018. Springer.

Nitzan, M., et al. (2005). Effects of external pressure on

arteries distal to the cuff during sphygmomanometry.

IEEE Transactions on Biomedical Engineering, 52(6),

1120-1127.

Nitzan, M., Patron, A., Glik, Z. & Weiss, A. T. (2009).

Automatic noninvasive measurement of systolic blood

pressure using photoplethysmography. Biomedical

Engineering Online, 8(1), 1-8.

Nitzan, M., et al. (2013). Comparison of systolic blood

pressure values obtained by photoplethysmography and

by Korotkoff sounds. Sensors, 13(11), 14797-14812.

Stergiou, G. S. & Bliziotis, I. A. (2011). Home blood

pressure monitoring in the diagnosis and treatment of

hypertension: a systematic review. American Journal of

Hypertension, 24(2), 123-134.

Stergiou, G. S., et al. (2018). Home blood pressure

monitoring in the 21st century. The Journal of Clinical

Hypertension, 20(7), 1116-1121.

Tholl, U., Forstner, K. & Anlauf, M. (2004). Measuring

blood pressure: pitfalls and recommendations.

Nephrology Dialysis Transplantation, 19(4), 766-770.

van Popele, N. M., et al. (2000). Arterial stiffness as

underlying mechanism of disagreement between an

oscillometric blood pressure monitor and a

sphygmomanometer. Hypertension, 36(4), 484-488.

Zheng, D., Pan, F. & Murray, A. (2013). Effect of

mechanical behaviour of the brachial artery on blood

pressure measurement during both cuff inflation and

cuff deflation. Blood Pressure Monitoring, 18(5), 265-

271.

A New Method to Determine Systolic Blood Pressure Indirectly Aided by Parallel Recording of ECG and PPG

227