A piezo-film-based measurement system for global haemodynamic assessment

IOP PUBLISHING

PHYSIOLOGICAL MEASUREMENT

Physiol. Meas. 31 (2010) 697–714

doi:10.1088/0967-3334/31/5/007

A piezo-film-based measurement system for global haemodynamic assessment Fabrizio Clemente1 , Pasquale Arpaia2 and Pasquale Cimmino1 1 Istituto di Ingegneria, Biomedica Consiglio Nazionale delle Ricerch, via Salaria km 29,300-00016 Monterotondo S. (RM), Italy 2 Dip.to di Ingegneria, Universit` a del Sannio, Corso Garibaldi 107, 82100 Benevento, Italy

E-mail: [email protected], [email protected] and [email protected]

Received 4 September 2009, accepted for publication 9 March 2010 Published 16 April 2010 Online at stacks.iop.org/PM/31/697 Abstract A non-invasive piezo-film-based measurement method for haemodynamic assessment is proposed. The design of a system, able to reconstruct the blood pressure waveform online by dealing with problems arising from the piezo-film capacitive nature in the targeted frequency range (from quasi-dc up to 12 Hz), is illustrated. The system is based on a commercial piezo-film placed easily on the radial artery with a special brace without any discomfort for the patient. The analogical conditioning circuit and digital signal processing are continuously tuned with the signal from the sensor to reconstruct the blood pressure signal online. Diagnostic schema, based on physio-pathological models, have been implemented in order to compute online trends of max[dP(t)/d(t)] and volemic status highly useful for the intensivist and anaesthesiologist. The system was characterized by numerical simulation and experimental in vivo comparison to the traditional reference system. Keywords: cardiovascular measure, piezoelectric sensor, non-invasive monitoring, pressure pulse (Some figures in this article are in colour only in the electronic version)

1. Introduction Continuous monitoring and analysis of physiological signals as well as online interactive signal processing are essential in the management of critically ill patients. In fact patient monitoring, and consequent interpretation of data, allows information on the patient’s status and on the therapy effects to be obtained (Milnor 1989, Nichols and O’Rourke 2005). In particular, 0967-3334/10/050697+18$30.00

© 2010 Institute of Physics and Engineering in Medicine

Printed in the UK

697

698

F Clemente et al

the arterial pressure waveform has received considerable attention for the assessment of cardiovascular systems for critical patients (Nichols 2005, Giomarelli et al 2004). The development of methodologies able to extract further information from blood pressure (BP) signals can reduce the global invasiveness of clinical measurement systems (Marino 1998, Clemente et al 2002, 2004, Giomarelli et al 2004, Papaioannou et al 2004, Mukkamala et al 2006) which is a key factor in the management of critical patients (Soncini et al 2002, Clemente et al 2004). Different applications show how different diagnostic data can be derived from the pressure waveform signal through simple computation by improving diagnostic capabilities and the ability to control therapies (Chen et al 1997, Clemente et al 2002, 2004, Papaioannou et al 2004, Giomarelli et al 2004, Mukkamala et al 2006, De Hert et al 2006, Tartiere et al 2007). Direct BP monitoring through the invasive arterial catheter is presently considered as the most accurate method in commercially available devices (Picco plus©, LiDCOTM plus, VigileoTM ). However, being invasive, this methodology cannot be applied in the ambulatory study and, moreover, turns out to be affected by various drawbacks, such as patient discomfort, the need for skilled professionals (for catheter insertion) and possibility of clinical complications. Currently, non-invasive, convenient and reliable measurement of the arterial pressure waveform is highly desirable for medical staff. Research activity is therefore more and more focused on non-invasive pressure monitoring from the peripheral artery (radial artery) and on consequent synthesis of systemic circulation characteristics (Chen et al 1997, Fetics et al 1999, Harada et al 2002, McLaughlin et al 2003, Tanaka et al 2005, Foo and Lim 2006). Derived from traditional wrist palpation, arterial applanation tonometry is one of the widely proposed methodologies for non-invasive BP measurement. It applies the principle of Imbert–Fick (Pressman and Newgard 1963): theoretically, the internal pressure of a spherical body with an infinitely thin, dried and elastic wall equals the force exerted by an external tangently applied plate on the body divided by the applanation surface area. Starting from the mechanical model of superficial artery applanation over an underlying bone (Pressman and Newgard 1963), in particular conditions, the displacement measured on the skin due to the pressure of radial artery pressed against the radial bone by an external bearing is directly proportional to the pressure itself. Therefore, since 1963 (Pressman and Newgard 1963) it has been widely studied and proposed for cardiovascular clinical application (Matthys 2004, Matthys et al 2008, Sorvoja et al 2005). As a matter of fact, the theoretical model to be transferred in the clinical device requires the sensing area to be smaller than the radial artery radius (few mm). Usually, the mechanical transducer is formed by a rigid holed applanation plate holding a displacement sensor of 2.5 mm2 area (i.e. a strain gauge). R Instrument) was based on a Therefore, the first commercial available system (Millar° micro transducer and turned out to be difficult to handle, even by skilled operators. Some R new commercially available devices (i.e. Colin° CBM 7000) exploit a sensor array in order to create perfect conditions for tonometry measurement. Of course, this kind of device requires multiplexing electronics, combined with an external computer unit having adequate computational capability. In practice, to date, the methodology shows problems in sensor positioning, as well as motion artefact (Matthys 2004), not yet solved. Furthermore, some of them seem to be phased out (Matthys et al 2008). These problems have been addressed by innovative measurement methods, based on collars (Tanaka et al 2005) or using different sensor technologies (Salo et al 2004, Gibbs and Asada 2005). A different methodology applies mathematical models to signals derived from distal arteries (Finapres Medical System Technology) but very frequent calibration, up to once per minute (Birch and Morris 2003, Schutte et al 2004), is required. However, until now,

A piezo-film-based measurement system for global haemodynamic assessment

699

none of them has given rise to a practical application in clinical monitoring and the debate on reliable non-invasive methodologies to measure the pressure waveform is still open (Birch and Morris 2003, Matthys et al 2008). In this paper, a non-invasive measurement method based on low-cost piezo-film to (i) monitor continuously BP at radial artery without any discomfort for patient and convenient for medical staff and (ii) implement new diagnostic tools for a non-invasive global haemodynamic assessment system useful in ICU and surgery is presented. In particular, in section 2, an outline of the proposed measurement procedure is given. In section 3.1, the piezo-film, the mounting system to place the sensor, the analogical conditioning circuit and the digital hardware are presented. In section 3.2, the software to autocalibrate the measurement system and to compute the radial artery pressure, from which further parameters are derived, is described. Finally, in section 4, the results of numerical simulation and in vivo experimental tests in order to (i) verify the reliability of the reconstructed pressure waveform and (ii) evaluate the drift in calibration during the time are reported. 2. The proposed measurement method In the following subsections, the background (section 2.1), the measurement principle (section 2.2), the diagnostic schema (section 2.3) and the procedure of the proposed method (section 2.4) are illustrated. 2.1. Background A piezo-electric element is able to convert force or pressure applied to its surface into a measurable electrical signal (Measurement Specialties 2006). The element is composed of two close conductive plates (electrodes), spaced by a polarized fluoropolymer, polyvinylidene fluoride (PVDF) film. From an electrical point of view, the sensor can be represented by a capacitor (figure 1), whose value C is affected by the PVDF dielectric constant. Vs is the piezo-electric generator voltage, directly proportional to the applied mechanical stimulus. When the voltage output of the sensor (Vo) is connected to a resistive load, usually tens of kÄ (Measurement Specialties 2006), the equivalent circuit acts as a high-pass filter. The cut-off frequency depends on the dimension of the PVDF film, according to the formula C = ε A/t, where A is the active film electrode’s area, t is the film thickness, and ε = εo ∗ εr (with εo the space permittivity and εr the relative PVDF permittivity) (Measurement Specialties 2006). The principle of blood pressure monitoring by means of extravascular sensors is based on measurement of the displacement of the tissue surface due to the force of the pressure variation. In practical devices, the pressure signal ranges from quasi-dc to 12 Hz. For avoiding information loss, a suitable cut-off frequency of the high-pass filter requires small dimensions for the sensor, by giving rise to positioning problems over the artery analogous to that in classical applanation tonometry. 2.2. Measurement principle Starting from these considerations, the intrinsic derivative characteristic at lower cut-off frequencies of the piezo-film was exploited for the proposed measurement method. In this range, a sensor of dimensions wider than the arterial radius measures the variation of the stimulus, i.e. the pure derivative of the pressure waveform. The acquired input signal represents the derivative of the pressure waveform (piezo-electric effect), and thus a signal

700

F Clemente et al

Figure 1. Electrical equivalent circuit of a piezo-film.

integration must be performed. Moreover, the sensor does not provide absolute pressure values; therefore, calibration for the pressure waveform based on gain (A) and offset tuning (B) is needed. These data can be obtained from values of systolic—max[Pas(t)]c—and diastolic— min[Pas(t)]c—pressures measured by different methods (normally by an oscillometric parallel device). Considering Pas(t) the radial artery pressure waveform, dPas(t)/dt is the output signal of the piezo-film sensor and the output pressure waveform from the presented procedure, Pasr(t), is Z dPas(t) dt + B, (1) Pasr (t) = A dt where A and B are computed from the calibration parameters (max[Pas(t)]c and min[Pas(t)]c) and the maximum and minimum of the integral of the acquired signal: Z dPas(t) I (t) = dt. (2) dt Therefore, the gain A turns out to be A=

max[Pas(t)]c − min[Pas(t)]c . max[I (t)] − min[I (t)]

(3)

Regarding the offset B, the difference between the two maxima, or the two minima is suitable. Owing to nonlinear behaviour, the average value is evaluated: abs{max[Pas(t)]c − A ∗ max[I (t)]} + abs{min[Pas(t)]c − A ∗ min[I (t)]} . (4) 2 From (3) and (4), A and B are continuously digitally adjusted to compute Pasr(t) according to (1). B=

2.3. Diagnostic schema Single continuous monitoring of the arterial pressure waveform is highly desirable for medical staff. As a matter of fact, the final goal of this paper is design of a system for use in computing derived haemodynamic parameters from non-invasive measurement of radial arterial pressure too. However, innovative parameters, even if limited and to be further crossed with other

A piezo-film-based measurement system for global haemodynamic assessment

701

haemodynamic data (i.e. pre-load, heart rate, etc), can be useful for new surgery procedures as well as at the bedside. Cardiac output (CO) and volemic status are the main parameters for the evaluation and maintenance of cardio-circulatory function in order to support the decision of the anaesthesiologist and intensivist (Marino 1998, Clemente et al 2002, Soncini et al 2002, Mukkamala et al 2006). Therefore, two algorithms for determining the trend of the (i) max[dPas(t)/dt] and (ii) systolic pressure variations (SPV) were considered. 2.3.1. max[dPas(t)/dt]. Elsewhere in particular circulatory conditions, the measure of the time derivative of peripheral arterial pressure, dPas(t)/dt, was shown to be correlated directly to the variation of CO (Clemente et al 2004, De Hert et al 2006) and to the left ventricular function (Tartiere et al 2007). 2.3.2. Systolic pressure variations (SPV). The effect of breathing on the systolic pressure (i.e. the maximum value of Pas) in mechanically ventilated patients was shown to have a valuable diagnostic capability for the volemic status, i.e. the volume of blood in the vascular system (Perel 1998, Clemente et al 2002, Soncini et al 2002). 2.4. Procedure The proposed measurement procedure was organized in the following steps (figure 2). (1) Sensing: the pressure exerted on the skin due to the radial artery, Pas(t), is measured by the sensor; according to the measurement method, this gives the time derivative of the arterial pressure signal. (2) Analogue conditioning: the sensor signal is shifted and amplified in order to exploit the characteristics of the analogue-to-digital (AD) conversion. A low-pass filter is required to avoid aliazing effects. (3) Digitization: AD conversion allows the signal to be treated according to the signal processing procedure and to implement diagnostic schema. (4) Calibration: an appropriate digital band-pass filtering allows the pressure signal to be determined from its time derivative and information loss to be avoided in the useful frequency range. Then, I(t) is computed according to (2) in order to compute Pasr(t) according to (3) and (4). Finally, Pasr(t) is presented continuously in real time to medical staff, in order to be used for standard physiological monitoring and for applying diagnostic schema. (5) Medical decision: according to the trend of global monitoring, medical staff supervise the patients’ haemodynamic condition and, on demand or periodically, perform the calibration procedure by measuring off-line systolic pressure, max [Pas(t)]c, and diastolic pressure, min [Pas(t)]c using the parallel oscillometric technique. Regarding the latter, as arterial cannulation is periodically checked (calibration, flashing, catheter displacement, etc) based on the awareness of medical staff (Marino 1998, Leung and Romson 2004), ongoing calibration is required here if the device parameters are in manifest conflict to the patient condition (i.e. due to mechanical displacement). 3. Measurement system design In the following subsections, details on the hardware (section 3.1) and the software (section 3.2) of the system designed according to the above measurement principle are given.

702

F Clemente et al

Figure 2. Flow chart of the implemented procedure.

3.1. Hardware The measurement system (figure 3) is mainly based on a sensor (section 3.1.1), a cuff device (section 3.1.2), a conditioning circuit (section 3.1.3) and (3.3.4) an acquisition section (section 3.1.4). 3.1.1. Sensor. The sensor is a piezo-film EMFi DT1–028 K of Measurement Specialties (figure 4), proposed elsewhere for pulse velocity assessment (McLaughlin et al 2003), owing to its capability of measuring the pulse pressure derivative. It was selected in the proposed system design owing to (Measurement Specialties 2006) (a) its high sensitivity to low-level mechanical movements, (b) its electrostatic shield located on both sides of the element (to minimize 50/60 Hz ac line interference), (c) its response to low-frequency movements in the 0.7–12 Hz range of interest and (d) its foil size suitable to the application (16 × 41 mm). In the development of the conditioning circuit, a value of equivalent capacitance of 1.3 nF was considered (Measurement Specialties 2006).

A piezo-film-based measurement system for global haemodynamic assessment

703

Figure 3. The proposed system, including brace, pneumatic cuff and PC.

Figure 4. Piezo-film sensor TD1-028 K (Measurement Specialties)

3.1.2. Cuff-device. The force exerted on the piezo-film pulse sensor placed on extravascular tissues depends on the hold-down pressure applied to the tissue surface. Thus, a suitable brace has been realized in order to position the sensor correctly. It holds up the arm, the wrist, the thumb and the sensor. An automatic inflation device was implemented in order to exert a force reducing noise effects due to micro-mechanical arm movements and gather the proper signal. R The developed system keeps the hand in the right measurement position with a LEXAN° ° R brace with Velcro strips and the sensor on the radial artery properly. From systolic, max[Pas(t)]c, and diastolic, min[Pas(t)]c, pressures measured with the classical oscillometric method, the cuff is settled to the best measurement pressure value, Pstop. This value must be enough high to have an acceptable signal without an excessive banding of the vessel wall during the measurement and range from 40 to 80 mmHg. 3.1.3. Conditioning circuit. The conditioning circuit (figure 5) was implemented in order to ensure easy detection of the signal from the transducer. The acquisition data system is

704

F Clemente et al

Figure 5. Schematic of the custom interface for analogue signal conditioning.

A piezo-film-based measurement system for global haemodynamic assessment

705

provided by an anti-aliasing filter and a gain stage in order to offer to the ADC a proper analogue input signal, with no data information losses (Bendat and Piersol 1986). The realized circuit carries out (a) the signal shift-up of 1.667 V in order to make it suitable for the microcontroller’s ADC too (U1A-R1-R2); (b) the sensor signal input buffering for impedance adaptation (U1B); (c) the filtering, implemented with the first order R7C9 node with 17 Hz cut-off frequency, in order to obtain only the bandwidth containing the pressure information (less than 15 Hz) and the ADC anti-aliasing effect; (d) the signal amplification, implemented with the high-performance instrumentation amplifier analogue devices AD623, which provides the desired high CMMR (>100 dB) and a gain of 6.5 (R8). As said above, the sensor circuit is a capacitor and a critical part of the interface circuit is the load resistance R6. It affects the frequency range of the derivative system as well as the signal amplitude. A 10 MÄ resistor is suggested (Measurement Specialties 2006) as the best compromise between (a) derivative quality, in order to obtain the real pressure waveform after the digital integration, and (b) frequency response, to get the whole pressure waveform bandwidth considered in the range up to 12 Hz. 3.1.4. Acquisition section. A PC-based system was selected for its capability to be easily substituted with a medical tablet PC in further engineering development. Currently, the acquiring system is constituted by (i) a high-performance battery-powered notebook PC (Pentium IV 3.2 GHz, 2 Gbyte RAM) in order to allow suitable digital signal processing algorithms to be experimented, and (ii) an acquisition board PCMCIA DAQ Card AI-16E4 (National InstrumentTM , Austin, TX). This configuration allows up to four signals to be acquired: one devoted to the signal from piezo-film and the others from different possible devices, in order to obtain a reference pressure waveform during in vivo test measurements too or for future expansions. The final version of the system considers the acquisition of a single channel derived from piezo-film. In the design, connections have to be carried out according to electrical safety practice, because connecting an industrial PC to medical devices can cause a loss of protection for patients, operators and the surrounding area. Therefore, the system has been developed according to safety rules imposed by the IEC 60601 standards’ family for electrical medical systems. Electrical safety tests were performed using the electrical safety tester for medical devices BIOTEK INC 601 SXL. 3.2. Software R The software (specially the user interfaces) was designed and implemented in LabVIEW° according to the standard of traditional medical devices, in order to allow an easy use by medical staff. In the following subsections, details about data acquisition (section 3.2.1), signal processing (section 3.2) and monitoring critical haemodynamic conditions (section 3.2.3) are given.

R 3.2.1. Data acquisition. The acquisition software is based on the functions of LabVIEW° library AI Config, AI Start, AI Read, and AI Clear, as well as on error chaining and the general error handler VIs. The acquisition is based on a classical double-buffered technique: data are continuously acquired into a circular buffer by using direct memory access (DMA) (Clemente et al 2000), at a sampling rate of 200 S s−1 owing to the 15 Hz pressure waveform bandwidth. The program allocates a buffer of 20 000 scans. At the acquisition start, ten scans are read R from the acquisition buffer on each external loop iteration and stored in a LabVIEW° array

706

F Clemente et al

to be treated. This setting guarantees Nyquist bandwidth and gives enough time and data to perform online scan-to-scan elaboration. When data are retrieved from the buffer, in each loop iteration, they are stored in a circular queue capable of containing up to 20 s of data. This signal time window is enough for the process algorithms to retrieve the relevant information. 3.2.2. Signal processing. The acquired input signal represents the derivative of the pressure waveform (piezo-electric effect), and thus a signal integration is performed. To obtain the best result, the dc offset effect is cancelled by using a 200 mHz Chebyshev filter. The resulting signal is integrated with a Rieman integration algorithm. A continuous calibration allows the effects of temperature on the PVDF dielectric constant to be cancelled (Measurement Specialties 2006) and mechanical displacement of the measurement system. Therefore, the calibration can be executed again during the running of the software on demand. 3.2.3. Monitoring critical haemodynamic conditions. The variation of max[dPas(t)/dt] is computed as the mean value over 20 s during the monitoring time (Anderson and Lyons 2001, Clemente et al 2000). The system performs SPV computation and its component according to the literature. After the calibration procedure, the value max [dPas(t)/dt] over a limited time in apnea conditions (without mechanical ventilation) is computed from acquired data and stored. SPV and its diagnostic components are computed and shown on the front panel when patient is mechanically ventilated (Perel 1998). 3.2.4. Data presentation. In figure 6, the implemented graphic user interfaces (GUIs) are shown. The goal is to provide an intuitive and easy-to-use instrument interface to users. Pressure, trends of max[dPas(t)/dt] and SPV in time domain and all the evaluated haemodynamic parameters are shown (figure 6(a)). The procedure for their computation is reported elsewhere (Clemente et al 2000). By using a proper window tab, the users can observe the signal coming from the measurement system hardware (figure 6(b)). The same GUI allows pressure waveform calibration by the input of new max[Pas(t)]c and min[Pas(t)]c values. 4. Device characterization The proposed system was characterized by numerical simulation (section 4.1), and in vivo tests (section 4.2). 4.1. Numerical simulation The system was verified via PSPICE simulation by using the following procedure. The in vivo signal acquired invasively is used as the input of a PSPICE circuit composed of (i) the electric equivalent of the used piezo-film according to the sensor’s Technical Manual (Measurement Specialties 2006), and (ii) the above-described analogue conditioning circuit. Therefore, the output of numerical simulation represents the input of the AD converter. The error of the sensor and conditioning circuit was assessed by comparing the simulation output with the digital time derivative of an in vivo signal acquired invasively (figure 7). The comparison confirms an error less than 4%.

A piezo-film-based measurement system for global haemodynamic assessment

707

(a)

(b)

Figure 6. Haemodynamic measurement interactive system synoptic. (a) Main monitor: upper trace—reconstructed pressure signal; lower trace: trend of computed parameters max[dPas(t)/dt] and SPV. Right side of the panel: numerical values of computed parameters. (b) Secondary monitor: upper side: calibration window; lower side: acquired signal trace.

4.2. In vivo tests Two procedures were used to (i) validate the estimated pressure waveform from the proposed instrument and (ii) verify the drift in the calibration critical in clinical non-invasive blood pressure measurement (Birch and Morris 2003, AAMI 2002).

708

F Clemente et al

Figure 7. Comparison between the analytical derivative pressure waveform (♦) and the simulated (PSPICE) output signal from the TD1-028 K sensor and the conditioning circuit using the digitized real pressure waveform as input (¤). The y-axis is normalized.

Table 1. Patient data and comparison between the intra-arterial and reconstructed pressure waveform.

Patient

Age (years)

Systolic HR (BTS pressure min−1) (mmHg)

1

73

80

115

57

2

80

85

102

49

3

80

81

101

46

Distolic pressure (mmHg)

Surgery

Mean error (invasive piezo) (mmHg)

Abd. aortic aneurysm Femoral −1.86 bypass Femoral by pass

SD (invasive piezo) (mmHg)

RMS (invasive piezo) (mmHg)

7.88

8.10

4.2.1. Pressure waveform. The developed system was validated in comparison to a reference system by in vivo measurements on three patients (table 1, column 1–6), monitored invasively in ICU in spontaneous breathing. Informed consent was obtained before patient admission. Patients are in ICU after different cardiovascular surgery for standard postoperative monitoring and treatment and are haemodinamically stable during the test. It is to be noted that the third patient showed typical post-operative tremors and unconscious movements. As reference, the monitoring system for invasive measurements—Hewlett Packard 1175 A Merlin—was used. Arterial pressure (Pas) was acquired invasively by using 20 G needle inserted into left radial artery. Non-invasive measurement was performed on the right radial artery with the proposed system which, in this case, also acquired the invasive signal simultaneously. Calibration, according to section 2.2, was here performed using systolic and

A piezo-film-based measurement system for global haemodynamic assessment

709

Figure 8. Upside invasive (- - -) and piezo-film-based instrument (——) pressure waveform. Underside their difference (invasive − piezo-film)

diastolic pressure data from the monitor. This is because an oscillometric device causes blood flow arrest in the cannulated artery. The record length of measurement for the three patients are, respectively, 19, 10 and 19 cardiac cycles covering at least one breath cycle. In figure 8, for one of the test cases, comparison between invasive and non-invasive pressure waveforms as well as the mathematical difference (error) between the acquired signals is reported. The capability of the proposed system to follow the variations of systolic pressure due to breathing is highlighted. The metrological performance of the proposed technique was estimated by assuming as a reference the invasive measurement system. Thus, the difference between the reference and piezo-film-derived pressure waveforms was computed for the three patients to verify the magnitude of disagreement. Data are reported for the three monitored patients (about 8000 samples) in the Bland–Altman plot (figure 9). Table 1 (columns 7, 8 and 9) reports mean error, standard deviation (SD) and RMS for all acquired samples in the three patients between invasive and piezo-film derived pressure values. 4.2.2. Drift in calibration. As the measurement principle is based on initial calibration from systolic and diastolic pressures measured with an alternative instrument, the drift in calibration over the time directly affects reliability of the computed beat-to-beat pressure waveform. Moreover these parameters are critical for arterial monitor (Zorn et al 1997, Marino 1998, Birch and Morris 2003) and in the proposed diagnostic schema. Therefore a deep error analysis was performed to assess the reliability of systolic and diastolic computed values in the use of the proposed method. The prototype system was validated in comparison to an oscillometric reference system (Omron M2 Compact HEM-7102-E by Omron Healthcare Co., Ltd). This approach has been applied in other studies to validate continuous beat-to-beat non-invasive blood pressure monitor (Zorn et al 1997, Schutte et al 2004) and is accepted by Dubl Educational Trust (Dabl Educational Trust 2010). The study was performed

710

F Clemente et al

Figure 9. Bland–Altman plots of the difference between the invasive and piezo-film-based instrument pressure waveform for the three monitored patients.

Table 2. Data of the 11 volunteers (9 male and 2 female).

Mean SD

Age (years)

HR (BTS min−1)

Systolic pressure (mmHg)

Disatolic pressure (mmHg)

38 13

75 7

117 15

70 25

on 11 healthy volunteers of different ages and sex, normotensive, with any heart diseases (table 2) recruited from university students and colleagues. Volunteers were informed on the measurement procedure and gave consent. Subjects were monitored for 15 min with the non-invasive automated oscillometric sphygmomanometer at brachial artery and the proposed system at the opposite arm. During measurements subjects were in a sitting position with both arms laid on a table at the level of the heart. Six measurements were performed at time intervals of 3 min. The first measurement was used to calibrate the proposed system according to section 2.2. In five following measurements, the values of systolic and diastolic pressures were simultaneously read from the oscillometric sphygmomanometer and from the proposed instrument. The difference (error) between values measured by two instruments was computed. Data were compared using Bland–Altman plots for both systolic (figure 10(a)) and diastolic (figure 10(b)) pressure values (Zorn et al 1997, Altman and Bland 1983, Schutte et al 2004). The mean difference, SD and RMS of the error between all systolic and diastolic pressure values are reported in table 3.

5. Discussion The aim of study is to propose a new measurement technique to monitor the non-invasively arterial pulse wave useful for anaesthesiological application to provide a reliable alternative for invasive measurements and for deep ambulatory cardiovascular monitoring.

A piezo-film-based measurement system for global haemodynamic assessment

711

(a)

(b)

Figure 10. Bland–Altman plots of the difference between the oscillometric and piezo-film-based instrument for (a) systolic pressure and (b) diastolic pressure. Table 3. Comparison between systolic and diastolic pressure measurement from oscillometric and the piezo-film-based devices.

Mean difference SD RMS

Systolic pressure (mmHg)

Diastolic pressure (mmHg)

1.02 6.60 6.67

−2.73 6.73 7.2

The first step of the work was to design and prototype a system requiring any complicated procedures or lengthy preparation. For this, the use of a piezo-electric film of much larger area than the vessel lumen has been proposed. The sensor was blocked to the arm by a rigid brace and a cuff device slightly pressing the film on skin at the artery level. Therefore, the signal from the sensor derives from the force exerted on the patient’s skin due to changes in pressure inside the vessel as in the traditional palpation method of peripheral arteries. To get the waveform from the sensor signal, a HW/SW system was designed and implemented. Since the device is based on the characteristics of the sensor electrical equivalent, a first assessment on the input signal to the AD converter was performed. This study shows good reliability of the project and acceptable error (error