Design, Development and Comparison of two Different Measurements Devices for Time-Resolved Determination of Phase Shifts of Bioimpedances
DDesign, Development and Comparison of twoDifferent Measurement Devices for Time-ResolvedDetermination of Phase Shifts of Bioimpedances
R. Kusche, S. Kaufmann, and M. Ryschka
Abstract —Bioimpedance measurements are a non-invasivemethod to determine the composition of organic tissue. Formeasuring the complex bioimpedance between two electrodes, analternating current with a constant amplitude is injected into thetissue. The developed voltage drop is used to calculate the real andimaginary part of the impedance under test. Measurements inthe past indicated that it could be possible that the beating of theheart has an effect on the measured phase shift of the impedanceunder test. In this work a hardware system is developed, capableof measuring changes of the bioimpedance phase shift with highresolution. Two different measurement methods are used. Thefirst method is an analog circuit, which has been developed ina previous project. The other method uses the integrated circuitAFE4300 (Texas Instruments) [1]. The results of both methodsare transmitted with a microcontroller (Xmega256A3BU fromAtmel Corp.) via the USB interface to a host PC. On the hostPC the visualization of the measurements and the control ofthe embedded system is achieved with a developed LabViewprogram. The work includes the development of the hardwareand software, as well as a comparison of the accuracy of the twomeasurement methods. The result of a verification of the systemsis that both measurement methods have an effective resolutionof about 0.3 ◦ . I. I
NTRODUCTION
Bioimpedance measurements are a non-invasive method fordetermining the electrical characteristics of organic tissues [2].With time-resolved measurements, it is possible to determinedynamic characteristics like the pulse wave or the effect ofbreathing [3].For measuring the magnitude and phase of the unknownbioimpedance, a small known alternating current is injectedinto the tissue. This current produces a voltage drop acrossthe bioimpedance, which is measured and used to calculatethe complex bioimpedance. Usually the measured impedancephase of organic tissues is negative. This behavior can be ex-plained by the capacitive behavior caused by the cell structureof biological matter [4]. Fig. 1 shows a simplified electricalequivalent circuit, whereby R represents the extracellularfluid and R and C represent the cell impedance with cellmembrane and intracellular fluid. When the bioimpedance is R. Kusche is with the Laboratory for Medical Electronics, L¨ubeck Univer-sity of Applied Sciences (telephone:+ 49 451 300 5400, e-mail:[email protected]).S. Kaufmann is with the Laboratory for Medical Electronics, L¨ubeckUniversity of Applied Sciences (telephone:+ 49 451 300 5400, e-mail:[email protected]).M. Ryschka is with the Laboratory for Medical Electronics, L¨ubeck Univer-sity of Applied Sciences (telephone:+ 49 451 300 5026, e-mail: [email protected]). R C R Fig. 1. Simplified electrical equivalent circuit of biological tissues measured time resolved a small variation of the impedancemagnitude can be observed which is synchronous with theheart beat.The reason for this effect is the pulsation of the blood in thetissue under test. When the heart is pumping blood throughthe bloodvessels, the pressure in the arteries increases abruptlyand then falls again. Because of this variation of pressure,the arteries diameter and thus the volume of blood in thetissue under test changes. Since the conductivity of blood issignificantly higher than that of tissue, this leads to a changeof the impedance. One may assume that besides the magnitudealso the phase of the complex bioimpedance is influenced bythe blood pulsation. The changing of the bioimpedance phaseshift would indicate that the blood additionally has a differentphase shift than the tissue. Typical bioimpedance phase shiftsof living tissues are between ≈ − ◦ to ◦ , with a maximumat about kHz [5].This work compares two different phase shift measurementmethods. The first method is an analog circuit, realizedwith Operational Amplifiers (OPA). The second method isimplemented by the IC (Integrated Circuit) AFE4300 fromTexas Instruments. For the evaluation of the different mea-surement methods, both methods are realized together withan Electrocardiography (ECG) circuit, as a physiological timereference, on the same Printed Circuit Board (PCB). The PCBalso contains a microcontroller system for Analog to DigitalConversion (ADC), basic signal preprocessing and USB fordata transmission to a host PC for further signal processingand displaying of the measurement data.II. M EASURING M ETHODS
A. Analog Measurement System
The basic idea of the analog phase shift measurement circuitis the subtraction of two alternating voltages with the samefrequency and amplitude. It can be shown that for smallphase shifts the difference of two equal sinusoidal voltages is a r X i v : . [ phy s i c s . i n s - d e t ] A ug lmost linear to their phase shift [6]. To apply this methodto determine the phase shift between the injection current,represented by the voltage drop across a shunt resistor andthe voltage drop across the bioimpedance, both voltages haveto be equalized in amplitude. In the actual implementation,this is achieved via normalization. Afterwards the differencesignal is rectified and low pass filtered to get a DC signal. Thecircuit has been realized within a previous project [6]. B. Integrated Measuring System (AFE4300)
The AFE4300 is an IC including analog measurementcircuits as well as a SPI bus for the communication with a mi-crocontroller. The interface between the analog and the digitalpart is internally realized with an ADC ( bit , SP S ) anda Digital to Analog Converter (DAC, bit , M SP S ). Thusthe IC is an integrated measurement system, there are differentkinds of signal chains integrated. In this work only the bodycomposition meter in I/Q (In-phase/Quadrature)-Demodulatormode is used. The necessary sinusoidal excitation current isalso generated inside the integrated circuit with a VoltageControlled Current Source (VCCS).III. H
ARDWARE D EVELOPMENT
The block diagram in Fig. 2 shows the schema of thedeveloped PCB for interfacing the object under test to thehost PC. For the communication between the AFE4300 andthe microcontroller, a SPI bus is used. However the externalanalog phase shift measurement circuit uses the PCB’s analoginputs.
Analog Phaseshift Measuring CircuitOptionalAnalogInputs
Microcontroller
ADCECG-Circuit-5 VAnalog 5 VAnalog3.3 VDigitalDC/DC-Converter LDO-Regulator4x OPAINA Integrated Phaseshift Measuring Circuit (AFE4300)
ATxmega256A3BU USB-UART-IC C o nn e c t o r S D - C a r d - S l o t S i g n a l - L E D s P u s h - B u tt o n s SPISPIADC S P I D i g i t a l O u t D i g i t a l I n P D I / J T A G UARTUARTUSB
PCB
Power SupplyAnalog Inputs Communication InterfaceAdditional Periphery
USB Port BUSB Port AINA
CH0CH1
ADCADCADC
Fig. 2. Block diagram of the realized PCB
A. Analog Inputs
To interface the analog phase shift measurement circuitan Instrumentation Amplifier (INA, LT1789-1 from LinearTechnology) is used. The results are fed to one channel ofa bit , kSP S dual channel ADC (LTC1865L fromLinear Technology). The second ADC channel is prepared for future use, together with four further analog inputs whichdigitalization is provided by the micro controller’s internal bit , M SP S
ADCs. Signal buffering is achieved byvoltage followers realized with OPAs (OPA4134 from TexasInstruments). The external high accuracy ADC is connectedvia SPI to the microcontroller.
B. ECG Circuit
For common mode reduction, the ECG module is equippedwith a Driven Right Leg (DRL) circuit and active drivenshields. It is based on an INA circuit (LT1789-1 from LinearTechnology) with baseline wandering rejection. In order toreduce noise an optional active 2 nd order lowpass filter and apassive Hz Twin-T notch filter are implemented [7].
C. Power Supply
The system is externally supplied with V DC , whichcould be provided from the USB. Though for electrical safetyconsiderations it is recommended to power the embeddedsystem with an IEC-60601-1 compliant power source and toisolate the USB via a fiber optic USB hub. Internally thesystem needs ± . V for the analog parts and . V forthe digital components. Whereas the . V are generatedvia a LDO-regulator (LT1129 from Linear Technology), the − . V are generated via a DC/DC-Converter (LT3479 fromLinear Technology). The converter’s switching frequency iskept above . M Hz to be far away from the measurementfrequency ranges. The positive as well as the negative supplyvoltages are filtered by analog low pass filters with a cut offfrequency of ≈ kHz . D. Microcontroller System and Communication Interfaces
The used microcontroller (ATxMega256A3BU from AtmelCorp.) is an bit controller with kB of program memory.It was chosen based on its peripheral features like integratedADCs and the USB-interface. The microcontroller has a clockfrequency of M Hz . For future use, further hardwarecomponents are implemented on the PCB like a SD-Card slot,signal LEDs and push buttons.Fig. 2 shows that the system has two USB ports (PortA & Port B). The difference between them is that Port Bdirectly connects the microcontroller with the PC. This isthe standard port for measurements. Port A connects themicrocontroller and the PC via an interface IC (FT2232D fromFuture Technology Devices Inc.). Port A is only used for debugpurposes. It is possible to program the microcontroller via thePDI or the JTAG interface also.Fig. 3 shows the manufactured and populated four layerPCB of the measurement system. It contains more than 300components and has dimensions of about mm x mm .IV. S OFTWARE D EVELOPMENT
The development of the software can be separated intotwo parts. The first part is programming the microcontroller’sfirmware. In the second part the LabView Software for the PCis programmed. ig. 3. Manufactured phase shift measurement PCB.
A. Microcontroller Firmware
For the firmware development the Atmel Software Frame-work (ASF) is used as an aid. Configuring the microcon-troller’s USB-Port as a Virtual Serial Port simplified thecommunication with the PC. Fig. 4 shows the flow chart of thefirmware. The first steps execute the hardware configurations
Microcontroller ConfigurationSTARTInterface InitialisationsRead from USBParameter received?Parameter=eParameter=rParameter=o Enable cyclic call of Subroutine 1
YESNO
Enable cyclic call of Subroutine 2Parameter=aParameter=d Disable cyclic calls
NONONONO YESYESYESYESYES
AFE4300 Initialisation NOSubroutine 1: · Measuring of Phase Shifts and ECG · Send Data with Timestamp via USB to PC Subroutine 2: · Measuring of Phase Shifts and ECG · Read all ADCs · Measure time of complete Subroutine Cycle · Send Data with Timestamp via USB to PC
Fig. 4. Simplified flow chart of the developed microcontroller firmware. and initializations. In the main loop the USB-Interface ischecked for new received data. The data packet sent from thePC consists of only one character. After a case distinctionthe respective measuring mode is executed. Subroutine 1 just measures phase shifts and sends the results to the PC.Subroutine 2 additionally measures all the other analog inputs.So Subroutine 1 is faster and the measuring cycle is shorter( T Sub ≈ . ms , T Sub ≈ . ms ). B. Labview Software
The programmed LabView software’s purpose is displayingthe measurements in real-time. It is also possible to changethe sample rates and activate or deactivate the ADC-channels.Additionally the results can be exported to an MATLAB orExcel compatible data format. To reduce the influence of noisea moving average filter can be activated. In Fig. 5 a screenshotof the software’s user interface is shown.
Fig. 5. Screenshot of the developed LabView software.
V. V
ERIFICATION AND M EASUREMENTS
For the verification of the phase shift measurements anequivalent circuit for bioimpedances according to Fig. 6 isused. The result of an error estimation, executed with thetotal differential, is that this circuit is more insensitive to thecomponents tolerances than the circuit shown in Fig. 1. C p R p R s Fig. 6. Equivalent circuit for bioimpedances used for verification
To realize the desired magnitudes and phase shifts only theresistors were changed. The reason is the higher tolerance ofthe capacitance. It is 1% whereas the resistors tolerance is just0.1%. An additional advantage is that resistors are available ina wide range of values. The verification of both systems wasexecuted with a frequency of . kHz . An error estimationresulted to a maximal phase shift error of the equivalent circuitof less than ± . ◦ in the interesting range.The impedances were realized on a separate PCB whereswitches are used to change the resistances. Fig. 7 showsthe result of the analog phase shift measurement system.Achieving this characteristic was only possible by adaptingthe system’s shunt resistance to the magnitude of the measuredmpedance. This measurement system shows a good linearitybut some offset depending on the magnitude of the measuredimpedance. Fig. 8 shows the results of the AFE4300. The -14-12-10-8-6-4-20-12,00 -10,00 -8,00 -6,00 -4,00 -2,00 0,00 M e a s u r e d P h a s e s h i f t / ° Real Phaseshift/° |Z|=50 Ω |Z|=400 Ω |Z|=700 Ω
Fig. 7. Measuring results of the analog system. The phantom’s phase shiftwas varied in a range of ◦ to − ◦ . Verification executed for magnitudesof
50 Ω ,
400 Ω and
700 Ω . measurements depend on the impedance’s magnitude, too.Furthermore the results have large offsets and are inverted.Especially for small phase shifts below ◦ there are strongdeviations from a linear relation. -45-40-35-30-25-20-15-10-12,00 -10,00 -8,00 -6,00 -4,00 -2,00 0,00 M e a s u r e d P h a s e s h i f t / ° Real Phaseshift/° |Z|=50 Ω |Z|=400 Ω |Z|=700 Ω
Fig. 8. Measurement results of the AFE4300. The phantom’s phase shiftwas varied in a range of ◦ to − ◦ . Verification executed for magnitudesof
50 Ω ,
400 Ω and
700 Ω . VI. R
ESULTS AND D ISCUSSION
The linear behavior of the analog measurement system,especially when there are small phase shifts, is an advantage.However the system’s major disadvantage is that an externalcurrent source is needed. For suitable results it is also impor-tant to adjust the shunt resistor. As a consequence the magni-tude of the impedance under test has to be known. The IC doesnot have such a linear behavior as the analog circuit. Althoughup to four calibration impedances can be connected to theIC pins. With these known impedances and a microcontrollerprogram it is possible to automate the calibration. The disad-vantage of using an IC with such a small housing (LQFP80)is the effort of implementation for experimental purposes. Acomplete PCB with a microcontroller has to be developedand manufactured. However there is an essential advantagethat the whole measurement is done by just one component.Thus no external current source and shunt resistor is needed.Furthermore the AFE4300 is low priced (2.50 $ per 1000 pcs.) and no additional ADC is required. For reliable measuringof absolute phase shifts both systems have to be calibratedcomprehensively. The effective resolution of both systems isabout . ◦ . Measurements of real bioimpedances could notexhibit a relationship between the pulse and the phase shift.Using another impedance measurement instrument, which hasalso been developed in the work group [8], [9], it could beshown that the blood pulsation induced phase shift changesare only in a range of about . ◦ . Hence for this applicationboth the analog and the integrated phase shift measurementsystems are inappropriate.VII. C ONCLUSIONS