Documente Academic
Documente Profesional
Documente Cultură
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
(http://iopscience.iop.org/0967-3334/34/1/1)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 193.145.230.6
The article was downloaded on 31/01/2013 at 08:20
E-mail: paco.bogonez@gmail.com
Abstract
This paper describes a multi-frequency single-channel electrical implantable
bioimpedance monitor (35 mm × 35 mm × 10 mm, weight 52 g) powered
by a NiMH battery. By using the tetrapolar method and injecting 10 μApeak,
the monitor is capable of measuring at 14 different frequencies, from 100 Hz
to 200 kHz. It contains a ZigBee transceiver to monitor the measurements
performed, and has an embedded memory for backing up the data. RC networks
and in-situ heart excised tissues were used to test the system. When measuring
a full spectrum every 5 min, 35 days of autonomy are possible due to the low
power consumption of the monitor. Temperature drift was estimated by short-
term and long-term measurements. Temperature cycling was used to measure
modulus and phase angle stability. The result was a very low effect on a modulus
decrease of 2.34 , with respect to an impedance of 322 , at 100 Hz and a
phase angle increase of 1.1◦ , at 200 kHz. In addition, measurement errors were
bigger at low frequencies because of the high impedance of the electrodes used,
which was higher than 10 k at frequencies below 1 kHz.
1. Introduction
The electrical bioimpedance technique has been revealed as a valuable tool in the diagnosis of
multiple diseases or tissular conditions, e.g. predicting the hypovolemia (Allison et al 2005,
Cai et al 2002), detecting the rejection of some transplanted organs (Parsonnet et al 2007),
Harms et al 2000, Harms et al 2001) or in the detection of tissular bubbles in divers (Jossinet
0967-3334/13/010001+16$33.00 © 2013 Institute of Physics and Engineering in Medicine Printed in the UK & the USA 1
2 P Bogónez-Franco et al
2. Methods
The study involving human subjects was approved by the Clinical Research Ethics Committee
of Fundació Institut de Recerca de l’Hospital Germans Trias i Pujol under the BELIC trial,
under the BELIC trial NCT00843245.
about 52 g. For powering the monitor, a NiMH battery with a capacity of 500 mAh was used.
Figure 2 shows the photograph of the monitor. All the measurement data are sent to the host
computer via RF, using the ZigBee protocol. In addition, as backup, measurements are stored
in a memory. The host computer controls the monitor behavior and displays the measurement
results.
reported impedance, in the order of 2 G from 100 Hz to 1 MHz. Such high value was achieved
using a generalized impedance converter (GIC) connected in parallel with the current source
with the aim of minimizing both, the output capacitance and the stray capacitance. The current
consumption of the Howland circuit alone is about 37.5 mW. Seoane et al (2006) designed a
current source based on an operational amplifier inverter driven by current instead of voltage.
The output impedance of this design is 150 k at 1 MHz.
With restrictions on size and power consumption, the voltage-controlled current–source
(VCCS) designed is based on the modified Howland circuit. The AD8038 was the operational
amplifier selected, due to its high bandwidth and low power consumption. Figure 3 shows
the circuit of the modified Howland current source. The maximum current of 10 μAp in the
range of 100 Hz to 200 kHz was injected to the tissue, being low enough respecting the limits
imposed by the IEC-60601. Figure 3 shows the scheme of the current source designed. Its
power consumption is of 5.6 mW.
Figure 5. Schematic representation of the input modification on the AD5933 for measurement
using the four-electrode method.
Biological tissues have much lower impedance than the 1 k (Gabriel and Gabriel 1997)
lower limit measurable by the AD5933. Due to the low impedance of biological tissues and
because the AD5933 adds a dc level, the current flowing through the tissue may exceed the
IEC-60601 safety limits. Furthermore the use of dc levels in voltage may cause burns and
electrolysis in the contact areas of electrodes. For measurements in biological tissues some
modifications in the injection/measurement stage must be carried out. Figure 4 shows the
block diagram of AD5933.
Seoane et al (2008) solved the problem previously mentioned by adding a voltage-to-
current converter using a VCCS to add a dc level. With these modifications the four-electrode
method was used for measurements. An additional operational amplifier was necessary for
using a voltage-to-current converter.
Due to the power restriction in the system designed, the voltage-to-current converter is
performed without any additional operational amplifier. In fact, the voltage sensed by the
instrumentation amplifier, which is proportional to the current flowing within the biological
tissue, is converted to a current using the op amp embedded in the AD5933. Figure 5 shows
the implemented voltage-to-current converter.
6 P Bogónez-Franco et al
2.7. Microcontroller
The bioimpedance monitor system is controlled by a microcontroller (MCU). For getting
monitor’s great autonomy, it becomes extremely important for every part of the monitor to
consume less power. For this reason the MSP430F2274 from the MSP430 family of Texas
Instruments microcontrollers was selected. The MSP430 was designed for low power devices,
having therefore a wide variety of operating power modes focused on reducing the power
consumption. The current consumption in mode LPM3 is 200 nA. Open-source Eclipse IDE
which is code limited to 16 kBytes, and low cost of the development tools were important
keys for selecting this microcontroller.
When turning on the system, the MCU configures I/O ports, serial ports and internal
variables and switching to the low power 32 kHz oscillator. Then, it is switched to LMP3
and the timer interrupts are enabled. To save power, the rest of the circuits are turned off.
Every second the MCU awakes and switches the oscillator to 1 MHz, turning on also the
power to the RF transceiver and waiting for incoming data during 250 ms, it gets back to the
low power state if no data are available. The system has been programmed to perform a new
set of measurements every 5 min. In the same period of time, the microcontroller switches
its internal oscillator to 1 MHz and turns on the analog part, performing measurements at
every frequency and storing them in the external memory. When the analog part is switched
off, the measured data are sent to the host PC. If no connection has been made between the
bioimpedance monitor and the host computer, data are re-sent for a number of programmed
times.
2.10. RF transceiver
The eZ430-RF2500 (Texas Instruments) was the transceiver selected which uses the 2.4 GHz
frequency band and has a current consumption of 21.2 mA when transmitting at 0 dBm
power level and 18.8 mA in the receiving mode. The maximum data transfer rate is 500 kbps.
The output power can be adjusted from −30 dBm to 0 dBm. In order to achieve a tradeoff
between range and autonomy, the data rate was set to 2400 bps and the RF power transmission
to −10 dBm. This development tool consists of two USB dongles to test the CC2500 RF
transceiver. There is one MSP430F2274 microcontroller and a CC2500 RF transceiver as parts
Implantable bioimpedance monitor using 2.45 GHz band for telemetry 7
of a USB dongle. It can be programmed using the USB dongle and it is possible to implement
different communication protocols like SimpliciTI, a reduced version of the ZigBee protocol.
The host PC is connected to a self-made board attached to a CC2500EMK (Texas
Instruments) Evaluation module. This evaluation modulus is connected to a seven element
Yagi antenna (model YAGI-7-2.4G from LPRS) with a gain of 9 dBi. The board also has
RS232 connection to interface with the PC.
2.11. Software
To control the different operating modes of the bioimpedance monitor, an application in Visual
Basic (Microsoft Corporation) was developed. By means of that application it is possible to
program the time during measurements, erase memory, read data stored in the memory,
program date and the time, perform only a single measurement and also read the received
signal strength indicator (RSSI) and the battery voltage. Additionally, the magnitude, the
phase, the temperature and the impedance locus graphs can be plotted.
2.12. Calibration
The bioimpedance monitor was calibrated using two saline solutions with conductivities
5.40 mScm and 89.3 μScm. These solutions were measured using a pH meter (pH meter
model Basic 20 from Crimson). The calibration was tested measuring over a saline solution
of 617 μScm.
3. Results
When the microcontroller is in sleep mode the current consumption is of 0.2 mA. In the
impedance measurement mode, the current consumption is 11.6 mA. For RF communication,
the current consumption is 34.1 mA and 2.7 mA for storing measurements in memory. Peaks
appear in the sleep mode because the RF transceiver is looking for new incoming data.
Experimentally the voltage of the battery was measured and at the end of the 30th day
the battery voltage had dropped 1 V, from 4.2 V to 3.2 V, indicating that there was yet enough
charge to power the monitor during some more additional days. In a period of 5 min, the
mean consumption calculated was 0.588 mA, having the bioimpedance monitor, therefore, a
theoretical autonomy of 35 days.
3.4. RC networks
The first network measured was formed by Rs = 51 , Rp = 150 and Cp = 100 nF, shown
in figure 9. The second network had the following values Rs = 68 , Rp = 130 and Cp =
100 nF.
The effect of electrode mismatch due to poor contact of the electrodes with the tissue is a
common cause of the errors in measurements (Bogónez-Franco et al 2009).
To model this effect, a PCB with two electrode impedances connected in series was created.
One of these electrode impedances can be short-circuited to cause an electrode impedance
mismatch of 100% in any of the leads of the bioimpedance monitor. Figure 13 shows the
electrical circuit used to cause the electrode impedance mismatch. As the final electrodes
that were to be used in the experiment were not ready at the time of writing this paper, the
skin–electrode impedance model was used for non-invasive EBI measurements.
For evaluating electrode effect, impedance was measured over a resistor of 320 with and
without a mismatch in each lead. It was observed that when a mismatch is provoked, electrode
impedance has a more noticeable effect when the mismatch is produced in the injecting leads
than when produced in the sensing ones, and also when frequencies are low instead of high.
10 P Bogónez-Franco et al
Figure 9. Impedance locus (imaginary part of the impedance versus real part of impedance) plot
of the RC network with the following values, Rs = 51 , Rp = 150 and Cp = 100 nF. Fluke
measurements (continuous line) and measured with the designed system (dotted line).
Figure 10. Modulus, phase and temperature measurements at 50 kHz on a 320 resistor during
5.5 h.
Figure 11. Modulus and phase measurement at 50 kHz on a 320 resistor under a temperature
profile.
Figure 12. Impedance plot of the lamb heart measured at the beginning, in 30, 50 and 90 min.
Solid line was the Cole function fitted with an algorithm created by the research group.
electrodes used to measure the heart excised tissue were custom-made stainless steel needles.
The injecting electrodes were located at the extremes, and the sensing electrodes in the middle
of the injecting electrodes. The distance between the injecting and sensing electrodes was
2.5 mm, and the distance between the injecting electrodes was 7.6 mm. The diameter of the
needles was 0.3 mm.
Dimensions of the lamb heart were: 60 mm long; 42 mm wide; weight 250 g. As the heart
came attached to the lungs, it was necessary to separate it. Afterwards the heart was put inside
a polyethylene box, which was then put inside a corrugated cardboard filled with small balls
of expanded polystyrene in order to create a thermal box. The heart temperature was 39.1 ◦ C
at the beginning of the measurements and 39.0 ◦ C at the end. The electrodes were placed in
the left ventricle. Figure 12 shows the Cole plot of the bioimpedance measurements.
Figure 13. Setup for PER measurement. The antenna was moved horizontally and vertically from
the center of the table.
Figure 14. Designed bioimpedance monitor before complete implantation inside the animal and
just before beginning PER measurement.
Table 1. PER error at different distances and heights from the receiver antenna.
Distance (m) Height (m) PER (%)
1 1.0 2.3
1.5 1.4
2.0 1.2
2 1.0 11.0
1.5 7.6
2.0 6.4
3 1.0 26.0
1.5 23.4
2.0 15.7
before discarding the communication. Even if data are not received, they are saved in the
internal memory.
The bioimpedance monitor will be implanted in the animal shoulder and placed
subcutaneously. This allows a reliable communication between the bioimpedance monitor
and the PC.
14 P Bogónez-Franco et al
4. Discussion
Electrode impedance mismatch has a net effect on the modulus impedance measurement,
although that effect is minimized because of the high input impedance of the sensing leads.
More significant errors in measurement are due to the limited output impedance of the
current source and the high electrode impedance at low frequencies. The proposed electrodes
used in the final measurements are built from a modified DSI Coupler Lead (Data Science
International), leaving the distal end of 5 mm uncoated with only one of the imbricate spirals.
Measurements errors at low frequency are related to the high electrode impedance and
probably, to the aliasing in the digital demodulation stage of the AD5933. To measure at such
low frequencies, the AD5933 must use a down-scaled clock.
Measurements of RC networks showed that the modulus largest error was produced below
2 kHz, being 2.77% the largest value, at 200 Hz. The phase error had a maximum value of
–51.66◦ at 100 Hz and fell up to −2.96◦ at 2 kHz. At higher frequencies the modulus error was
below −0.25% and the phase error remained below −0.8◦ . Those were the results because the
impedance of the RC network at low frequencies is in the order of some tenths of k and with
the limited output impedance of the current source it has a noticeable effect. Measurements
on biological tissues were planned to start at the end of July 2012.
There is more noticeable effect when the mismatch is produced in the injecting leads
than when produced in the sensing leads and when frequencies are low instead of high; all
this is related with the output impedance of the current source. For example, at 700 Hz with
no electrode mismatch, the modulus and phase errors are −3.43 and 0.02◦ , whereas when
the electrode mismatch is produced in the injecting leads, I+ and I−, the errors are −14.36
and −14.28 in the modulus and 1.37◦ and 1.20◦ in the phase, respectively. When the
mismatch is produced in the sensing leads, V+ and V−, the errors are −6.50 and 1.85
in the modulus and 0.40◦ and −0.23◦ in the phase, respectively. At low frequencies the
impedance of the electrode had a value of 10 k at 100 Hz and of 7 k at 1 kHz. In the
sensing leads this effect is less important due to the high input impedance, in the order of M
because of input buffers. Degradation in the CMRR of the detection circuit is greater.
No influence on measurement results were observed when doing short- and long-term tests.
A temperature increase in the designed system from 28 ◦ C to 40 ◦ C, produces a maximum
decrease of 2.34 in the modulus, in relation to an impedance of 320 at 100 Hz, and a
maximum increase in the measured phase of 1.10◦ at 200 kHz.
Using ZigBee Protocol to transmit data has been proved in a valid way to extract data from
inside the body. Valdastri et al (2008a) used three implantable devices to transmit pressure
and temperature from stomach with that protocol. The goals of that experiment were to record
the minimum transmission power levels from different parts of the gastrointestinal (GI) tract
and to compare those levels to international safety regulations. The power density emitted
by the implanted devices was measured with a strength meter, being the result 38 mW/m2,
much lower than the reference level taken from the International Commission on Non-Ionizing
Radiation (ICNIRP). To demonstrate the feasible data transmission in the GHz bands, Poon
et al (2007) performed some simulations making later measurements to check-out the results
Implantable bioimpedance monitor using 2.45 GHz band for telemetry 15
obtained. She assembled an antenna of 2 cm and the receiver was a 2 mm width coil, and used
beef sirloin for the measurements. Due to the un-optimized transmitter antenna the optimal
frequency for a 2 cm antenna was 1 GHz.
5. Conclusions
An electrical implantable impedance monitor was created and tested in RC networks and heart
excised tissue. It has small dimensions and RF communications can be used to extract data
generated during measurements. A memory is also included to back up data.
Power consumption is very low, allowing autonomy of 35 days when measuring every
5 min.
Short and long-term measurements were done for characterization of the monitor.
Temperature drifts were also measured, appreciating no great errors in modulus and phase
measurements.
Errors in measurements were greater at low frequencies due to the high impedance of
electrodes used and the low output impedance of the current source designed.
Preliminary experiment was carried on with animals and is shown in Sánchez et al (2012).
The experiment was performed on seven pigs wherein myocardial infarction was induced; two
of them were used as control. The results are coherent with the known behavior of normal,
acute ischemia and healed scar tissue. The bioimpedance monitor allows us to provide an
indicator to the transition between these tissue states.
Acknowledgments
The presented work was supported by grants SAF2008-05144-C02-02 from the Spanish
Ministry of Science and Innovation; 080331 from Fundació La Marató de TV3, Redes de
Investigación del Instituto de Salud Carlos III (REDINSCOR, RD06/0003) and Fondo Europeo
de Desarrollo Regional (FEDER).
We would like to thank Sı́lvia Núñez and José Núñez for their valuable help in the
correction of this document.
References
Akin T, Najafi K and Bradley R 1998 A wireless implantable multichannel digital neural recording system for a
micromachined sieve electrode IEEE J. Solid-State Circuits 33 109–18
Allison R D, Ray Lewis A, Liedtke R, Buchmeyer N D and Frank H 2005 Early identification of hypovolemia using
total body resistance measurements in long-term care facility residents Gend. Med. 2 19–34
Bertemes-Filho P, Brown B H and Wilson A J 2000 A comparison of modified Howland circuits as current generators
with current mirror type circuits Physiol. Meas. 21 1–6
Bogónez F, Vázquez F, Surakhy O, Sevilla J and Riu P 2006 Portable wireless bioimpedance measurement system
for haemodialysis monitoring Conf. Proc. World Congress on Medical Physics and Biomedical Engineering
(Seoul) pp 2238–41
Bogónez-Franco P, Nescolarde L, Bragós R, Rosell-Ferrer J and Yandiola I 2009 Measurement errors in
multifrequency bioelectrical impedance analyzers with and without impedance electrode mismatch Physiol.
Meas. 30 573–87
Cai Y, Zimmerman A, Ladefoged S and Secher N H 2002 Can haemodialysis-induced hypotension be predicted?
Nephron 92 582–88
Capstick J W and Wood T B 1922 The effect of change of temperature on the basal metabolism of swine J. Agric.
Sci. 12 257–68
Cook R D, Saulnier G J, Gisser D G, Goble J C, Newell J C and Isaacson D 1994 ACT 3: a high speed high precision
electrical impedance tomograph IEEE Trans. Biomed. Eng. 41 713–22
16 P Bogónez-Franco et al
Gabriel C and Gabriel S 1997 Compilation of the dielectric properties of body tissues at RF and microwave frequencies
appendix C: modelling of the data http://www.niremf.ifac.it/docs/DIELECTRIC/home.html
Grimnes S and Martinsen Ø 2000 Bioimpedance and Bioelectricity Basics (New York: Academic)
Harms J, Schneider A, Baumgartner M, Henke J and Busch R 2001 Diagnosing acute liver graft rejection: experimental
application of an implantable telemetric impedance device in native and transplanted porcine livers Biosens.
Bioelectron. 16 169–77
Harms J, Schneider A, Hess U, Pragst I, Henke J, Busch R and Böttcher K 2000 Telemetric impedance analysis
of the liver: evaluation of noninvasive device for diagnostics of acute graft rejection after experimental liver
transplantation Biomed Tech. 45 43–50
Jossinet J, Fourcade C and Gardette B 1981 An application of bioelectrical impedance for the detection of tissular
bubbles in the diver: preliminary results Med. Biol. Eng. Comput. 19 419–25
Orlic D, Kajstura J, Chimenti S, Limana F, Jakoniuk I, Quani F, Nadal-Ginard B, Bodine D M, Leri A and Anversa P
2001 Mobilized bone marrow cells repair the infarcted heart, improving function and survival Proc. Natl Acad.
Sci. USA 98 10344–4
Parsonnet V, Marak M J, Panken E, Zucker M J, Villanueva A, Kucher T, Driller J, Tuder G, Olesnicky L and Combs W
2007 Detection of early renal transplant rejection by minimally-invasive monitoring of impedance variability
Biosens. Bioelectron. 22 2749–53
Poon A, O’Driscoll S and Meng T 2007 Optimal operating frequency in wireless power transmission for implantable
devices Proc. 29th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBS 2007) pp 5673–8
Ross A S, Saulnier G J, Newell J C and Isaacson D 2003 Current source design for electrical impedance tomography
Physiol. Meas. 24 509–16
Sánchez B, Guasch A, Bogónez P, Gálvez C, Puig V, Prat C, Semino C, Bayes A and Bragós R 2012 Towards on line
monitoring the evolution of the myocardium infarction scar with an implantable electrical impedance spectrum
monitoring system Conf. Proc. Annu. Int. Conf. of the IEEE Eng. Med. Biol. Soc. EMBS ’12 pp 3223–6 at press
Seoane F, Bragós R and Lindecrantz K 2006 Current source for multifrequency broadband electrical bioimpedance
spectroscopy systems. A novel approach Proc. 28th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. pp 5121–5
Seoane F, Ferreira J, Sánchez J J and Bragós R 2008 An analog front-end enables electrical impedance spectroscopy
system on-chip for biomedical applications Physiol. Meas. 29 S267–78
Valdastri P, Menciassi A and Dario P 2008a Transmission power requirements for novel ZigBee implants in the
gastrointestinal tract IEEE Trans. Biomed. Eng. 55 1705–10
Valdastri P, Rossi S, Menciassi A, Lionetti V, Bernini F, Recchia F and Dario P 2008b An implantable
ZigBee ready telemetric platform for in vivo monitoring of physiological parameters Sensors Actuators A:
Physical 142 369–78
Zábaco O 2010 Informe anatomopatológico Universidad de Santiago de Compostela. Santiago de Compostela, Spain