A REAL TIME CARDIAC MONITORING SYSTEM

Arterial Pressure Waveform Capture and Analysis

V. G. Almeida, T. Pereira, E. Borges, J. M. R. Cardoso, C. Correia

Instrumentation Centre, Physics Department, University of Coimbra, R Larga, Coimbra, Portugal

H. C. Pereira

Instrumentation Centre, Physics Department, University of Coimbra, R Larga, Coimbra, Portugal

ISA- Intelligent Sensing Anywhere, Coimbra, Portugal

Keywords: PIC microcontrollers, dsPIC, Arterial pressure waveform, Physiological signals, Cardiac system, Embedded

systems, Real time.

Abstract: An arterial pressure waveform recorder and analyser based on a Microchip PIC microcontroller (µC),

dsPIC33FJ256GP710 is described in this article. Our purpose is to develop a dsPIC based signal monitoring

and processing system for cardiovascular studies, specially dedicated to arterial pressure waveform (APW)

capture. We developed a piezoelectric (PZ) probe designed to reproduce the APW from the pulsatile activity

taken non-invasively at the vicinity of a superficial artery. The advantages in developing a microcontroller

based system show up in decreasing the associate cost, as well as in increasing the functionality of the

system. Based on a MathWorks Simulink platform, the system supports the development and transfer of

program code from a personal computer to the microcontroller, and evaluation of its execution on rapid

prototyping hardware. Results demonstrate that embedded system can be an alternative to be used in

autonomous cardiovascular probes. Although additional studies are still required, this probe seems to be a

valid, low cost and easy to use alternative to expensive and hard to manipulate devices in the market.

1 INTRODUCTION

The social and economic impact of cardiovascular

diseases and the importance of efficient early

diagnostic tools keep mobilizing the interest of many

researchers (Laurent et al., 2006). Continuous

monitoring and analysis of physiological signals, as

well as online interactive signal processing are

essential in the management of ill patients. The term

arterial stiffness denotes alterations in the

mechanical properties of arteries, as the decay of

elasticity in the arterial wall fibers. Much effort has

focused in determining the best way to measure this

parameter: pulse pressure, pulse waveform analysis

and pulse wave velocity (PWV) measurements are

some examples.

Historically the cuff sphygmomanometer was the

first method to quantify a part of the medical

information contained in the arterial pressure

waveform (APW); however it provides a limited

amount of information: quantitative blood pressure

information at two specific points of the APW.

Electrocardiography (ECG) is another widely

accepted method to extract cardiovascular

information but is rather limited when arterial

stiffness information is concerned.

The APW morphology has gained clinical

interest due the additional information obtained from

the time-varying pulse waveform (Avolio et al.,

2010), such as the pattern of the ventricular ejection

and the elastic properties of the arterial tree.

Ideally, we are looking for an instrument capable

of delivering the calibrated, precise APW at the

ascending aorta, even though from a remote sensing

site (peripheral artery). Non-invasive assessment of

APW typically uses waveforms recorded at one of

two anatomical locations: the radial and the carotid

artery. Carotid blood pressure is often used as a

surrogate for central aortic blood pressure due its

location. Van Bortel et al. (2001) showed that the

carotid pulse pressure differ only 1.8 mmHg from

central aortic pulse pressure.

G. Almeida V., Pereira T., Borges E., M. R. Cardoso J., Correia C. and C. Pereira H. (2011).

A REAL TIME CARDIAC MONITORING SYSTEM - Arterial Pressure Waveform Capture and Analysis.

In Proceedings of the 1st International Conference on Pervasive and Embedded Computing and Communication Systems, pages 83-90

DOI: 10.5220/0003369600830090

 SciTePress

In this work we concentrate in developing a non-

invasive device suitable to carotid APW

measurements and to further processing, from which

a great deal of clinically relevant information can be

derived. The resulting instrument assumes the shape

of a real time system, autonomous, with minimal

human intervention, capable to respond to the time

variations of the physiological signals.

Real time embedded systems using digital signal

processors (DSP) in biomedical applications

assumed, over the last years, an increasing

importance due to the enhanced functionalities that

they are capable of imparting. The development of

this technology has enabled significant

improvements in speed of analysis, accuracy, noise

immunity, programmability, size reduction and, in

addition, a decrease in cost.

Numerous cardiovascular applications have been

reported in the literature: Klig et al. (1978) uses

these systems for monitoring blood pressure and

ECG signals. Bing-Nan et al. (2004) proposes an

embedded medical advisory system for mobile

cardiovascular monitoring devices that provides

microcirculation information. Germano et al. (2009)

introduces a generic architecture for developing

biomedical embedded systems with special

application for clinical analysis and for patient

monitoring.

The Explorer 16 development board with its

attached microcontroller is used with some

additional hardware in order to configure a fully

operational system.

The real time operating system is discussed

along the paper, as well as the details of data

acquisition, data pre-processing and data

transmission to the host computer. In Section 2 a

general embedded system design is briefly

introduced while the software parts are described in

Section 3. In Section 4, experimental results are

shown demonstrating a very good overall

performance in an almost autonomous (minimum

human intervention) mode of operation.

2 EMBEDDED SYSTEM DESIGN

The microcontroller (µC) was selected from the

Microchip PIC family due to its features and

embedded resources. These µCs are widely available

on the market at relatively affordable prices.

Moreover, a wide range of programming tools are

also available (Bansal et al, 2009, Smolnikar and

Mohorcic 2008).

The dsPICs are a hybrid solution that combines

the processing power of a DSP with the functionality

of a microcontroller, which includes fast interrupt

vectors, control of peripherals, general purpose I/O

and can run compact code.

The dsPIC33 family, in particular, employs a

powerful 16-bit architecture that integrates the

control features of a microcontroller with the

computational capabilities of a DSP. The

dsPIC33FJ256GP710 was chosen due to its

characteristics: 40 MIPS processor speed, 256 kbyte

program memory and 30 kbyte of RAM.

The Explorer 16 development board (figure 1 a)

is a low cost, efficient development board to

evaluate the features and performance of

Microchip's Microcontrollers, in particular the

PIC24FJ128GA010 and the dsPIC33FJ256GP710.

Top and bottom views of a piezoelectric (PZ)

probe responsible for capturing the APW at the

carotid artery site, are shown in figure 1 b) and c).

The architecture of the system is

diagrammatically represented in figure 2.

Figure 1: a) Explore 16 development board (a) and PZ

probe, top and bottom views, respectively b) and c).

The signal acquisition/processing block is

responsible for amplifying the sensor signal and

identifying the positive peak (one of the prominent

points of the APW).

The signal conditioning block has the function of

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

Figure 2: Workflow of the system.

supplying unipolar (positive only) values as required

by the dsPIC ADC. The microcontroller is

programmed to deliver the final pressure waveform

and transmit data via RS232 to a Graphical User

Interface (GUI) on MATLAB.

The PZ probe and rs232gui are discussed in

detail in the remaining of this section.

The USB (Universal Serial Bus) is also

responsible for supplying power to the system.

2.1 PZ Probe

PZ based probes have been widely used in APW

measurements along the last years as a result of their

characteristics: high sensitivity, high signal-to-noise

ratio SNR, as well the low price associated.

A PZ element is able to convert force or pressure

applied to its surface into a measurable voltage

signal. From an electrical point of view the sensor

can be modelled as an AC coupled voltage generator

(Karki, 2000) and, consequently, it does not respond

to static excitation.

Due to the above mentioned electrical

characteristics of the PZ probe, the collected signal

appears as a time derivative of the APW that excites

the PZ sensor.

In figure 3 a) the configuration of the probe is

shown. The PZ sensor (2) in use is the MURATA

7BB-12-9 sounder, this is attached on a double

printed circuit board (PCB) (3).The interface

between the transducer and artery (or silicon tube) is

done by a PVC piece (1) (in form of a “mushroom”,

with 15 mm diameter in top). The probe’s covering

consist in a plastic box (OKW (ENCLOSURES)-

B9002107).The final belt-mounted sensor used to

carotid artery in vivo acquisitions is shown in figure

3 b).

The principle of APW measurement is based on

the transmission of its mechanical energy that shows

up as a displacement of the tissue surface (carotid

artery) to the PZ surface.

Figure 3: a) The PZ probe configuration, the arrow

indicates the externally applied forces, (1) mushroom-

shaped interface, (2) PZ disc sensor and (3) printed circuit

board (PCB) and in b) the final probe is shown.

Figure 4 shows a typical response of the PZ

(gray line) to an APW-like excitation (black line)



A REAL TIME CARDIAC MONITORING SYSTEM - Arterial Pressure Waveform Capture and Analysis

obtained in a dedicated test bench (Pereira et al.,

2009). As the microcontroller cannot sample

negative voltages, the PZ signal is level shifted by a

convenient DC value before being fed to the

amplifier. Figure 5 depicts the level shift circuit.

Typically, if collected at the carotid site, it shows

peak amplitudes of around 1V and exhibits a SNR in

the order of 40 dB, allowing subsequent signal

processing algorithms to run free from noise induced

errors.

Figure 4: PZ signal (gray line) in response to APW

excitation (black line). Blue line represents the DC value.

Figure 5: Level Shifter circuit used to PZ signal, before

being fed to the ADC.

2.2 User Interface

As mentioned previously, data are uploaded via

RS232 interface. The system is capable of sending

captured data in real time with a signal acquisition

rate high enough to be useful in real time

hemodynamic monitoring, sampling rate of 1kHz.

This program reads the serial RS232 port and

displays the data in a graph. It stores the received

data in an individual text file for each measurement.

This GUI is based on the one developed by Kerhuel

(2010).

3 SOFTWARE MODULES

In the system design, the speed of computation and

memory capacity are considered as top importance

characteristics.

Microchip MPLAB v8.30 is used for building the

modules using the C30 compiler for C

programming, which simplifies code generation.

We also use the MathWorks Simulink platform

to generate C code. It provides an interactive

graphical environment in which the algorithms are

developed in the form of block diagrams. With the

aid of Real Time Workshop Embedded Coder it can

be used to generate the target independent ANSI C

code. The code generated can be included into

MPLAB IDE projects. The flowchart below

represents the programming stages. The program is

based in the available blockset Embedded Target for

PIC/dsPIC (Kerhuel, 2010).

Figure 6: Flowchart of the stages of programming. The

microcontroller transmits the data in real time through

UART (universal asynchronous receiver/transmitter).

The C code includes the following software

modules:

• Integration function

• Peak identification

• Baseline restoration

• Time Delay

0 0.2 0.4 0.6 0.8 1

-0.5

0.5

1.5

U1A

TL082CD

INA128P

VCC

VSS

VCC

VSS

PZ signal (LS)

PZ signal

Time (s)

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

3.1 Physiological Assumptions

We start by recalling that, due to the inherently

capacitive nature of the sensor, its voltage signal will

inevitably occurs in the shape of a time derivative of

the APW and, consequently, the output stage will

perform an integration of the signal in order to

recover the original APW.

The prominent point’s identification is an

important task to identify different phases of the

cardiac cycle, and for hemodynamic parameters

extraction (Almeida et al, 2011).

As is well known, signal integration requires a

periodic reset signal to avoid saturation. A time

reference, for the reset signal must be identified in a

pulse by pulse basis. The prominent peak in figure 4

(gray line) corresponds to the highest rise slope of

the pressure waveform, few milliseconds after the

beginning of the pulse (ventricular contraction). To

identify this peak the PZ signal is fed to a peak

detector (figure 7) formed by two peak stretchers

(U3, D1, R4, C1 and U4, D2, R5, C2) and a

comparator (U5).

To avoid false triggers the peak stretchers have

different time constants, R4C1 and R5C2

, chosen

according to the typical signals, and one this is level

shifted before being fed to the comparator.

Figure 7: Peak detector circuit.

This mechanism contributes to the baseline

restoration process. The elimination of baseline drift

consists in forcing the foot of systolic pulses to start

close to zero without affecting the shape of the

signal.

Like many other biosignals, APW pulse drift

essentially correlates to three sources: respiratory

activity, variations in signal shape and signal jitter

(defined as a random variation of the of the pulse

period). If an operator holds the probe during data

collection, an extra source of baseline drifts shows

up due to the variations of its interaction with the

patient. In in vivo tests a collar is used to eliminate

the influence of the operator.

In a typical measuring session, the first few

pulses are used just to gather the two main system

adjustment parameters: time delay and baseline

level, both to be used in the signal integration. Then

real data collection starts for as long as possible (no

discomfort for the patient that is asked not to

swallow during data acquisition). Typically, one to

two minutes allows the acquisition of a number of

cardiac pulses high enough for the statistical

processing that follows.

In figure 8 a flowchart of the system design is

represented.

Figure 8: System overview. Acquisition/processing

platform of the system.

4 RESULTS

The real time embedded system developed for

cardiovascular applications was tested using cardiac

simulated waveforms, synthesized using a weighted

combination of exponential functions (Almeida et

al., 2010).

An Agilent 33220A arbitrary wave generator

delivers the signal that excites the system via a test

input coupled through a capacitor of the same order

of magnitude of the sensor capacity itself).

In vivo tests are performed in some volunteers

that granted their previous written, informed

consent.

Peak detection (Figure 9) and delayed PZ

waveform (Figure 10, black line) intermediate

signals are shown. The relative error is computed

from the data shown in Figure 11 a), comparing the

integrated waveform with the excitation waveform

(Agilent).

TL081CD

VCC

VSS

R4 C1

TL081CD

VCC

PZ signal

VSS

R5 C2

VCC

LM311D

B/STBVS+ BAL

VS-

5 6

VCC

VSS

VCC

Peak

detection

Level

Shifter

A REAL TIME CARDIAC MONITORING SYSTEM - Arterial Pressure Waveform Capture and Analysis

4.1 Peak Identification

The more prominent positive peak (green line) is

identified, as shown in the figure 9 using the circuit

described in figure 7.

As mentioned previously, this peak in the

differentiated signal (PZ diff) corresponds to the

highest rise slope of the arterial pressure waveform.

Figure 9: Results from the peak identification circuit.

4.2 Delay

The ideal delay value required, tmin, rising time of

the most prominent peak in the PZ signal, not being

possible to measure, so a time delay td is calculated

for this purpose, To determine td we use information

about systolic upstroke time, tu, of general APWs,

Buteler (1961) measures the change in systolic

upstroke time in patients with different diseases. The

upstroke time varied between 110 ms and 230 ms. In

our measurements a 110 ms time delay is used to

perform the integration.

min2 ttu ×≈

(1)

230110 ≤≤ tu

(2)

11555 ≤≤ td

(3)

Figure 10 shows the original PZ signal (gray

line) and the delayed PZ signal (black line).

Integration (blue line) is performed for the delayed

PZ signal, as is visible.

4.3 Integration

The system is capable of recovering the APW from a

PZ probe, as is shown in figure 11 where the

excitation (blue) and the recovered waveforms (light

blue) are shown.

Figure 11 b) plots the relative error (defined as

the difference in amplitude from original and

recovered waveform). Table 1 resumes the statistical

parameters of the measurements. The data are

Figure 10: PZ waveform is represented in gray line and

delayed PZ waveform in black line. The integrated

waveform PZ int (integrated) represented in blue line was

obtained from the delayed PZ waveform. tmin - ideal

delay value, tu- time during of systolic upstroke, td- delay

time used in the integration of the PZ waveform.

characterized by a mean value, a minimum and a

maximum, as well as by the standard deviation

(STD) for the relative error, between the excitation

and the recovered waveforms. Maximum error

occurs in the ascendant edge, 88 ms and minimum

error occurs at 450 ms. The mean value is only 2.19

%, and the. STD is 2.57 %.

Figure 11: a) Integrated waveform (PZ int) and the

excitation waveform (Agilent). b) The relative error

between the original and the final waveform.

0 0.5 1 1.5 2

-200

200

Time (ms )

0 0.5 1 1.5 2

0.5

diff

Peak detection

A.U.

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

0.5

Time (m s)

A.U.

probe

(delay)

PZ int

0 100 200 300 400 500 600 700

0.2

0.4

0.6

0.8

Time (ms )

A.U.

Agilent

PZ int

0 100 200 300 400 500 600 700

Time (ms)

| Agilent-PZ int | (%)

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

Table 1: Statistics information of measurements depicted

in figure 11 b).

Maximum

(%)

Minimum

(%)

Mean

(%)

STD Deviation

(%)

18.520 0.001 2.1999 2.572

4.4 In vivo tests

To assess the capability of the system in

distinguishing different points in APW morphology

a small universe of volunteers was analysed.

The PZ probe is held by a collar and placed in

the carotid artery site for in vivo data acquisition

(figure 12).

The output shows a typical waveform where the

most prominent points are easily identified: systolic

peak (SP), reflection point (RP), dicrotic notch (DN)

and dicrotic peak (DP). The effect of the baseline

restoration mechanism that prevents baseline

fluctuations along time can also be seen. In figure 13

a set of pulses, about 7 seconds are shown, where is

possible identify a typical morphology of the APW

with its prominent points identified in b).

Figure 12: Our probe is held by a collar and placed over

carotid artery.

Figure 13: a) A set of pulses and in b) is shown a detail

view of one pulse. RP-reflected point, SP-systolic peak,

DN-dicrotic notch, DP-dicrotic peak.

5 CONCLUSIONS

In this paper, we presented the design of a real time

cardiac monitoring system for APW capture. A PZ

sensor was integrated in a signal acquisition circuit

that communicates with a dsPIC. Application

software running on the matlab was also developed

to receive and plot APW signals.

Currently we are studying the clinical use of our

probe, in medical environment, comparing our data

with catheter collected data to prove that this system

is a valid alternative with low cost associated.

Algorithms for the patient’s information must be

integrated in order to extract information about:

heart rate, AI, PWV and reflection points (Almeida

et al 2010, Almeida et al 2011, Pereira et al, 2010).

The proposed system was designated with user-

friendly interfaces which easy the usability of this

system and reduce the need of a long-time-training

before usage. The minimal human intervention is a

fundamental characteristic for this purpose.

ACKNOWLEDGEMENTS

We acknowledge support from Fundação para a

Ciência e a Tecnologia for funding (PTDC/SAU-

BEB/100650/2008 and SFRH/BD/61356/2009) and

from ISA, Intelligent Sensing Anywhere.

REFERENCES

Almeida V., Pereira T., Borges E., Figueiras E., Cardoso

J., Correia C., Pereira H. C., Malaquias J. L. and

Simões J. B., 2010. Synthesized cardiac waveform in

the evaluation of augmentation index algorithms. In

IEEE EMB, Proceedings of the 3rd International Joint

Conference on Biomedical Engineering Systems and

Technologies (BIOSTEC 2010). Valencia, Spain 20-

23 January 2010.

Almeida V., Santos P, Figueiras E, Borges E, Pereira T,

Cardoso J, Correia C., 2011. Hemodynamic features

extraction from a new Arterial pressure waveform

Probe. Accepted to be presented in Biosignals 2011

(BIOSTEC 2011). Rome, Italy.

Avolio A. P., Butlin M., and Walsh A., 2010. Arterial

blood pressure measurement and pulse wave analysis -

their role in enhancing cardiovascular assessment.

Physiol. Meas. 31, R1–R47.

Bansal D., Khan M.,Salhan A. K., 2009. A real time

embedded set up based on digital signal controller for

detection of bio-signals using sensors. Sensors &

Transducers Journal, Vol. 105, Issue 6, pp. 26-32

0 1 2 3 4 5 6 7

0.5

Time (s )

0 0.2 0.4 0.6

0.2

0.4

0.6

0.8

Tim e

(

)

A REAL TIME CARDIAC MONITORING SYSTEM - Arterial Pressure Waveform Capture and Analysis

Bing-Nan L., Ming-Chui D., Vai M. I., Mak P. U., 2004.

An embedded medical advisory system for mobile

cardiovascular monitoring devices. IEEE International

workshop on biomedical circuits & systems. S2.1-1-4

Buteler, B., 1961. The relation of systolic upstroke time

and pulse pressure in aortic stenosis.

Germano J., Ramalho R., Sousa L., 2009. On the design of

distributed autonomous embedded systems for

biomedical applications. CST Pervasive Health 2009.

Karki J., Signal Conditioning Piezoelectric Sensors.

[Online] Texas Instruments, Application Report,

SLOA033A (2000) Available at: http://focus.ti.com

/lit/an/sloa033a/sloa033a.pdf [Accessed 21 June 2010]

Kerhuel L., Embedded Target for dsPIC. [Online]

Available at: www.kerhuel.eu. [Accessed 01

November 2010]

Klig V. 1978. Biomedical applications of microprocesors.

Proceedings of the IEEE, vol. 66, no.2, 151-161

Laurent S., Cockcroft J., Van Bortel L., Boutouyrie P.,

Giannattasio C., Hayoz D., Pannier B., Vlachopoulos

C., Wilkinson I., and Struijker-Boudier H., 2006.

Expert consensus document on arterial stiffness:

methodological issues and clinical applications.

European Heart Journal, 27, 2588–2605.

Mackenzie I. S., Wilkinson I. B. and Cockroft J. R., 2002.

Assessment of arterial stiffness in clinical practice. Q J

Med., 95, 67–74.

Pereira H. C., Cardoso J. M., Almeida V. G., Pereira T.,

Borges E., Figueiras E., Ferreira L. R., Simões J. and

Correia C., 2009. Programmable test bench for

hemodynamic studies IFMBE Proc. 25 1460ff

Pereira, H. C., Pereira, T., Almeida, V., Borges, E.,

Figueiras, E., Simões, J. B., Malaquias, J. L., Cardoso,

J. M. R., Correia, C. M. B., 2010. Characterization of

a double probe for local pulse wave velocity

assessment. Physiol. Meas 31,1449–1465

Smolnikar M., Mohorcic M., 2008. A framework for

developing a microchip PIC microcontroller based

applications. Wseas Transactions On Advances In

Engineering Education, vol 5, Issue 2, pp83-91

Van Bortel L. M., Balkestein E. J., van der Heijden-Spek

J., Vanmolkot F. H., Staessen J. A., Kragten J. A.,

Vredeveld J. W., Safar M. E., Struijker-Boudier H. A.

and Hoeks A. P., 2001. Non-invasive assessment of

local arterial pulse pressure:comparison of applanation

tonometry and echo-tracking. J. Hypertens. 19 1037–

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems