An amplitude to time conversion technique suitable for multi-channel data acquisition and bioimpedance imaging

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 7, NO. 3, JUNE 2013 349

An Amplitude-to-Time Conversion TechniqueSuitable for Multichannel Data Acquisition

and Bioimpedance ImagingJong Cheol Baeg, Student Member, IEEE, Hun Wi, Student Member, IEEE, Tong In Oh, Member, IEEE,

Alistair Lee McEwan, Senior Member, IEEE, and Eung Je Woo, Senior Member, IEEE

Abstract—In this paper we exploit the high timing resolutionoffered by microprocessors to develop an amplitude measurementapproach that is convenient for high channel count portable sinu-soidal recording systems such as the bioimpedance measurementsused in impedance imaging. This approach reduces the number ofcomponents required per channel, reducing cost, size and powerconsumption compared to the traditional approaches. The setupuses two high performance comparators to convert amplitudedifference to a timing difference. This is captured by a high speedmicroprocessor. A straightforward algorithm removes DC andtiming offsets. We suggest three modes of operation: fast: less thanone period of the input, normal: exactly one input period and highprecision: multiple input periods. The mean signal-to-noise ratiowas 40, 81, and 112.4 dB in fast, normal, and high precision moderespectively for a range of resistive loads.

Index Terms—Amplitude-to-time conversion, analog-to-digitalconverter (ADC), capture unit, bioimpedance imaging, micropro-cessors, multichannel data acquisition.

I. INTRODUCTION

T HE electrical bioimpedance spectrum is an inherent char-acteristic of biological tissue related to the intracellular

and extracellular volume and cell membranes. These changedue to the physiological and pathological status of tissuesso bioimpednace may be used as a powerful diagnostic tool.Bioimpedance is widely used in applications ranging from cellmonitoring [1]–[7] and physiology [8]–[10] as it is sensitive tochanges as small as the ion channels of the cell membrane to thelarge, temporary change in tissue structure during respirationor blood circulation.

Manuscript received March 24, 2012; revised June 13, 2012; accepted July19, 2012. Date of publication September 28, 2012; date of current version May22, 2013. This work was supported by a grant from Kyung Hee University in2012 (KHU-20120821). This paper was recommended by Associate EditorS. Leonhardt.J. C. Baeg, H. Wi, and E. J. Woo are with the Department of Biomedical

Engineering, College of Electronics and Information, Kyung Hee University,Yongin-si, Gyeonggi-do 446-701, Korea.A. L. McEwan is with Department of Biomedical Engineering, College of

Electronics and Information, Kyung Hee University, Yongin-si, Gyeonggi-do446-701, Korea, and also with the School of Electrical and Information Engi-neering, The University of Sydney, NSW 2006, Australia.T. I. Oh is with Impedance Imaging Research Center and the Department of

Biomedical Engineering, College of Electronics and Information, Kyung HeeUniversity, Yongin-si, Gyeonggi-do 446-701, Korea (e-mail: [email protected]).Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TBCAS.2012.2212437

In bioimpedance, a safe sinusoidal current at controllablefrequencies is applied to a pair of electrodes and the resultingvoltage changes are measured at additional electrodes [11].Noise reduction may be greater with bioimpedance thanstraightforward biopotential measurement as we can demodu-late at the known applied frequencies. In this paper we chosea 500 Hz frequency source to test the system as it is above thebandwidth of physiological signals ( Hz) and low enoughnot to be adversely effected by reactance ( MHz) [12]–[14].Imaging systems based on multichannel bioimpedance, such

as electrical impedance tomography (EIT) will benefit fromhigh channel counts but this increases the cost, size and powerconsumption [15]–[20]. The purpose of this paper is to describean amplitude-to-time conversion approach that can achieveefficient multichannel bioimpedance measurements.Many EIT research groups have used analog or digital

phase-sensitive demodulation techniques to detect the changeof impedance distribution inside the body [21]–[23]. Whilethese techniques are accurate, they are not suitable for highchannel count systems due to the high cost and calcula-tion overhead of the demodulation algorithm. They requireseveral analog components or multichannel high resolutionanalog-to-digital converters (ADCs) which are sensitive totemperature variations, mismatch, crosstalk, power supplynoise and reference voltage stability.Recently there has been an interest in developing ampli-

tude-to-time conversion for analog to digital conversion toparticularly address the variation effects in the deep sub micronCMOS processes that have been primarily developed for highspeed digital logic [24]–[26]. The advantage comes from ex-ploiting the high time resolution of high speed logic or clockingcircuits (typically a voltage controlled oscillator) to converttime difference signals that represent the original amplitudechanges. A counter is used to convert the time differencesto digital values [27]. This is a much more efficient processthan flash ADC that requires comparators, successiveapproximation (SAR) ADC that requires succession steps,and delta-sigma ADC that requires oversampling, filteringand negative feedback [28]–[33]. For biompedance and EITimaging we suggest an off the shelf implementation of ampli-tude-to-time conversion that is different from the monolithicdesigns and conventional ADC as it only uses two comparatorsand the capture unit of the high speed microprocessor.In this paper, we will describe the multichannel data acquisi-

tion method using a time difference for bioimpedance measure-

1932-4545/$31.00 © 2012 IEEE

350 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 7, NO. 3, JUNE 2013

Fig. 1. Block diagram for ATC.

ments. It is applicable to any measurement using the amplituderesponse to a sinusoid stimulus such as audio measurements,amplitude modulation or power measurements. There are 3 dif-ferent operating modes: fast acquisition mode, normal acqui-sition mode and high precision mode. There is a simple algo-rithm to convert the time difference to the amplitude of mea-sured signal. When a signal includes DC offset, we can cali-brate the measurement to remove the offset using a DC offsetnulling method. We apply comparator calibration because thecomparators are not ideal. We compared the performance ofa range of discrete comparators and operating modes by theirmean signal-to-noise ratio (SNR) on a serially connected re-sistor network.

II. METHODS

A. Amplitude-to-Time Converter (ATC)

Our method of amplitude-to-time conversion is based on theobservation that sinusoids of different amplitudes can be ob-served to intercept a delayed reference signal at different phasesor time delays, (1).

(1)

We are interested in only measuring the amplitude as we planto measure bioimpedance signals at low frequencies up to sev-eral kHz where phase information is minor. We also assume nodistortion or non-linearity of the sinusoid which is reasonablefor bioimpedance signals below a few volts when using standardAg/AgCl based electrocardiogram (ECG) electrodes applied tothe skin.

B. System Design

Fig. 1 shows a block diagram for the proposed system. Wefirst apply a sinusoidal current to the subject and measure theresulting voltage which is modulated by bioimpedance changesdue to e.g., respiration of the lungs increasing in size and fillingwith air. The voltage at each electrode is buffered by a dedicatedbuffer for each electrode then multiplexed to Comparator B inthe ‘sensing unit’.

A reference sinusoid is generated by delaying a copy of theapplied sinusoidal signal with a fixed phase in the ‘phase-delayunit’. Comparator B detects the intercepts or crossing points ofthe phase delayed reference and sensed voltage resulting in thesquare wave, Capture B. A second square wave, Capture A isgenerated by detecting the zero crossings of the phase delayedreference.The time difference between edges of Capture A and Capture

B is related to the amplitude of the sensed voltage as it repre-sents the time taken for the phase delayed reference to pass fromzero volts to the amplitude of the sensed voltage. We calculatethe estimated amplitude value from time differences in the ‘Am-plitude Calculation Unit’ using (2).

(2)

C. DC Offset Calibration

DC offsets are a common issue when using skin based elec-trodes due to the half cell potential that exists between metal andionic conductors. These might be removed by a filter but thatwould add an additional component with associated sensitivityto temperature variations and mismatch, particularly relating tothe phase for low frequencies close to DC. Digital phase sensi-tive demodulation techniques reject DC offsets but they requirehigh computation overheads.However the amplitude-to-time conversion technique pre-

sented here can easily remove DC offsets becuase they arerevealed in mismatch of the time difference between risingedges and falling edges in the Capture waveforms. The fol-lowing (3)–(5) describe the DC subtraction algorithm. Equation(3) describes the time difference between rising edges, T1. Herethe sensed voltage has an amplitude and DC offset .We have added our deliberate phase offset to the referencesinusoid which has a different DC offset . Equation (4)describes the time difference between falling edges. By takingthe difference between (3) and (4) we arrive at (5) and (6)shows that the amplitude can be recovered by the ratio.

(3)

(4)

(5)

(6)

D. Comparator Calibration

Non idealities in the comparators such as DC offsets,threshold voltage variations and effects of implementation suchas the PCB setup, reference source and phase delaying will leadto fixed amplitude errors. These were calibrated by measuringthe number of clock cycles counted by the capture unit withsources directly connected to the comparator, i.e., with a zeroohm load.

BAEG et al.: AMPLITUDE-TO-TIME CONVERSION TECHNIQUE SUITABLE FOR MULTICHANNEL DATA ACQUISITION 351

Fig. 2. (a) Waveforms in normal operating mode and high precision mode. Inthis example the amplitude of the sensed waveform is changed depending onmeasurement electrodes. Multiplexer was changed its configuration at 1, 5, and9 ms, respectively. T1 is the time between the positive to negative zero crossingof the phase delayed reference and when it crosses the positive value of thesensed voltage. T2 is the same for the negative case, recorded to remove DCoffset. T3 and T4 are the equivalent over 2 periods as an example of the high pre-cisionmode by averaging. T5 to T8 are again the equivalent for the sensedwave-form coming from the different measurement electrode and are visibly shorterthan T1 to T4. (b) Time converted waveforms following the comparators.

E. Operating Modes

The system can be operated in three operating modes thattrade-off precision for measurement speed.1) Normal Operating Mode: In the normal operating mode

each channel is converted over a single period. An example withamplitude decreasing is shown in Fig. 2(a). It can clearly beseen that the times between the rising edges of Capture A andB decrease (T5 is shorter than T1), Fig. 2(b).2) High Precision Mode: When the amplitude is constant

over more than one period we can increase our precision bycounting the increased time differences, T3 and T4 or T6 andT8 in Fig. 2(a) and (b). This is similar to averaging with theadvantage that the large, 32-bit counter in the microprocessormay be used to accumulate the time over multiple samples.3) Fast Acquisition Mode: In high channel count systems it

may be desired to multiplex several electrodes to each measure-ment channel. Typically in bioimpedance imaging systems thereare many voltage recording electrodes situated between currentinjection electrodes. The electric field decreases rapidly awayfrom the current injection electrodes. This can causes the ampli-tude of the measured voltages to rapidly decrease. These volt-ages depend mainly on electrode location, boundary shape andthe fixed tissue background so their range can be estimated priortomeasurements. Themonotonically increasing amplitudes cor-respond to increasing time difference between the crossing timepoint and the reference point so these may be allocated timemultiplexed positions within a single period. An example with

Fig. 3. (a) Waveforms in fast acquisition mode. Voltages measured at electrode1 and 2multiplexed to the sensedwaveform and phase delayed reference. T1 andT2 are the equivalent for the normal and high precision mode. i.e., T1 is the timebetween the negative going zero crossing of the phase delayed reference and thefirst time it crosses the positive sensed voltage (multiplexed to electrode 1). T3 isthe time between the negative going zero crossing of the phase delayed referenceand the next time it crosses the sensed voltage (multiplexed to electrode 2). T2and T4 are the equivalents for the negative voltages. After measured T1 andT2 for electrode 1, multiplexer changed its connection to electrode 2. (b) Fastacquisition mode time converted waveforms following the comparators. T1 isthe time between the rising edge of Capture A and the first rising edge of CaptureB, following the falling edge of Capture A. T3 is the same with the additionaltime to the next rising edge of Capture B. T2 and T4 are the same for the fallingedges of Capture A measured to the rising edges of Capture B, after Capture Arises.

two electrodes is shown in Fig. 3(a). T1 and T2 correspond tothe smaller amplitude of electrode 1. The sensed voltage is thenswitched to electrode 2 where T3 and T4 measure a longer timedelay from Capture A, due to the increased amplitude as shownin Fig. 3(b).

F. Implementation

We implemented the system with a high speed micropro-cessor (Delfino EVM board included TMS320C28346) runningat 300 MHz. The resolution of capture units is 1/(300 MHz),i.e., 3.333 nsec. The system relies on the performance of thecomparators so we built a custom PCB for testing a numberof comparators. In the sensing unit we used buffers on each of16 electrode inputs for high input impedance and a 16 channelmultiplexer (ADG1406, Analog Devices, USA). The operatingmodes, normal, precise or fast defined the operation of the mul-tiplexer. All 16 channels connected to different electrodes, re-spectively. In high precision mode the channels sample eachelectrode for more than one period (in normal mode) and in fastmode the multiplexer switches between electrodes within eachsource waveform period.The TMS320C28346 with 6 enhanced capture modules

allows the calculation of 6 crossing point times as the reference

352 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 7, NO. 3, JUNE 2013

time point is common. In our design they can work togetherto reduce the measurement time (fast mode) or increase theprecision (high precision mode). The capture interrupts aresensitive to any noise at the comparator outputs, therefore weoperate the capture units in one shot mode to escape the noisytransition periods when the multiplexer switches. We alsoused the input qualifier provided with the microprocessor as anon-linear digital noise rejection filter to suppress glitches.In the microprocessor capture unit, there is a 32 bit counter

running at 300 MHz in free running mode. When detecting arising or falling edge, the counter value is stored in a first-in-first-out (FIFO) buffer and can be used to generate an interrupt.Here, we used the edge detection of the signal Capture A (whichindicates the reference time point) to generate an interrupt toread the FIFO contents of both capture units.The microprocessor can communicate with the PC to transmit

data and receive commands for control of the ATC circuit. Weimplemented USB and Bluetooth communication methods in aPC communication module. The speed of USB and Bluetoothwere 930 kbps and 230 kbps, respectively.In initial tests we found that the system is highly sensitive to

power supply noise as this produces glitches at the comparatoroutputs. Therefore in all tests we used the most stable powersupply available to us. This was a power supply developed in thelab based on a medically isolated SwitchedMode Power Supply(ECM100US07, ECM100US09, XP Power Limited, Singapore)and precision regulators (TL431, Texas Instruments, USA) withthe following characteristics: Line regulation and load regula-tion error 0.05%. Ripple rejection ratio over 105 dB. Noise lessthan dB up to 100 kHz.

G. Experimental Setup

First, we evaluated the performance of three candidatecomparators, HA4905 (Intersil, USA), LM319 (FairchildSemiconductor, USA) and AD8564 (Analog Devices, USA) infast acquisition mode on a simple resistor phantom. The resistorphantom consisted of fifteen resistors with 1% tolerancesoldered together in a serial line. We applied a 2 , 500 Hzsine wave voltage to the source point and measured voltagesfrom each of the 16 measurement points using the proposedamplitude-to-time conversion method. These voltages weremeasured simultaneously using the 16-to-1 analog multiplexer,ADG1406. The reference waveform lagged 45 degrees behindthe source signal. Each time the comparator outputted a risingedge or falling edge, the microprocessor captured the time lagbetween the cross point and reference point, and calculatedthe difference as the amplitude. The percentage error from theknown impedance was calculated after applying comparatorcalibration and DC offset calibration. We compared the meanSNR over 512 repetitions. The SNR was measured as the meanrecorded amplitude divided by the standard deviation, (7).

(7)

The best performing comparator was compared with a 24-bitaudio sampling card and tested in all three modes: fast, normaland high precision over a range of resistor values.

TABLE ICOMPARATOR CALIBRATION RESULTS WITH 512 AVERAGES

III. RESULTS

In fast mode, the comparators had the following mean SNR:LM319 37.9 dB, AD8564 34.7 dB and HA4905 40 dB. For thethree comparators measured, the SNR varied by less than 0.5 dBover 512 repeated measurements.The comparator calibration results are shown in Table I for

500 Hz. The HA4905 provided the best performance whichmay be due to its low offset voltage, current and crosstalk.The LM319 is a widely used comparator without any specificcircuitry for high speed or low noise operation. The AD8564has very low latency of 7 ns which caused it to produce moreglitches than other comparators and reducing the measuredSNR. This was also evident from our calibration values wherethe falling edge was often missed, causing the calibrationvalue for the falling edge to be much larger than the othercomparators.We choose the HA4905 as the best comparator and tested

its performance against a 24 bit ADC. The best SNR of theATC with HA4905 was 112.4 dB over 256 periods which com-pared well with the 24 bit ADC which had an SNR of 113 dB.As expected the SNR of the ATC system with the HA4904comparator degraded gracefully with less periods: 32 periods96.1 dB, 8 periods 90.3 dB, 4 periods 87.2 dB, in normal mode(1 period, 81.4 dB) and in fast mode (40 dB).With 1600 measurement channels at 500 Hz in normal acqui-

sition mode, all measurements were finished in 0.533 seconds.In high precision mode, the measurement time will increase bythe number of successive periods measured and the SNR willbe improved.

IV. DISCUSSION AND CONCLUSION

We have presented an accurate amplitude measurementsystem that requires few components and is therefore attractivefor developing high channel count systems. The conventionaltechnique of using a high resolution ADC requires enoughsamples per period to satisfy the Nyquist sampling while thepresented ATC technique only requires two samples per period.For computation the ATC technique needs two subtract oper-ations and one division operation. This is efficient comparedto the conventional digital phase sensitive demodulation tech-nique which requires many more computations to achieve thesame result. In keeping the component count low, the techniquemay be robust to temperature and electromagnetic noise. Inthis paper we propose this technique only for the measurementof the magnitude of the impedance signal. It relies on a fixedphase in the ‘phase delay unit’ and has only been tested here fora sinusoidal sensed waveform with no distortion, nonlinearityor significant noise. These conditions may not exist in some

BAEG et al.: AMPLITUDE-TO-TIME CONVERSION TECHNIQUE SUITABLE FOR MULTICHANNEL DATA ACQUISITION 353

bioimpedance applications particularly where the sensed signalis very small or when there are artefacts due to electrode move-ments or changes in the contact impedance. At low frequencies( MHz) the permittivity of tissue is small so we can expectto keep a good relative phase for the phase reference signal.In this paper we chose a phase lag of 45 degrees for the phaseof reference signal. This was a compromise between the theo-retical best phase lag of 0 degrees which would be affected bynoise in the comparators and 90 degrees which will produce asmall time difference. In future work the relationship betweenthese variables needs to be investigated.The high precision stems from using the fine time resolu-

tion offered by the microprocessor and in presenting the am-plitude information as a time difference between two analogsignals. This subtraction process is able to remove DC offsetand decrease noise. Comparator calibration can compensate fornon-ideal characteristics of the comparator and source. We alsoexpect advantages in developing synchronous systems as wedo not rely on multiplexing many channels into a high perfor-mance ADC, rather we simplify the amplitudemeasurement andare able to dedicate one of these systems to a single or smallnumber of channels. The high precision mode result showed theexpected increase in SNR of approximately 6 dB for each pe-riod doubling.We have an interest in developing large channel count sys-

tems for a planer EIT system for breast cancer detection fromimpedance images that would feature 2400 channels and mi-croscopy systems that would require over 8000 channels [34],[35]. Clearly the complexity of the recording channel is veryimportant in this design.

REFERENCES

[1] Y. Huang, Y. Liu, B. L. Hassler, R. M. Worden, and A. J. Mason, “Aprotein-based electrochemical biosensor array platform for integratedmicrosystems,” IEEE Trans. Biomed. Circuits Syst., pp. 1–9, 2012,Early Access Article.

[2] J. Guo, J. Yuan, andM. Chan, “Modeling of the cell-electrode interfacenoise for microelectrode arrays,” IEEE Trans. Biomed. Circuits Syst.,pp. 1–9, 2012, Early Access Article.

[3] E. Sarro, M. Lecina, A. Fontova, C. Sola, F. Godia, J. J. Cairo, andR. Bragos, “Electrical impedance spectroscopy measurements using afour-electrode configuration improve on-line monitoring of cell con-centration in adherent animal cell cultures,” Biosens. Bioelectron., vol.31, no. 1, pp. 257–263, 2012.

[4] M. Vergani, M. Carminati, G. Ferrari, E. Landini, C. Caviglia, A.Heiskanen, C. Comminges, K. Zor, D. Sabourin, M. Dufva, M.Dimaki, R. Raiteri, U. Wollenberger, J. Emneus, and M. Sampietro,“Multichannel bipotentiostat integrated with a microfluidic platformfor electrochemical real-time monitoring of cell cultures,” IEEE Trans.Biomed. Circuits Syst., pp. 1–10, 2012, Early Access Article.

[5] L. Yao, P. Lamarche, N. Tawil, R. Khan, A. M. Aliakbar, M. H.Hassan, V. P. Chodavarapu, and R. Mandeville, “CMOS conducto-metric system for growth monitoring and sensing of bacteria,” IEEETrans. Biomed. Circuits Syst., vol. 5, no. 3, pp. 223–230, 2011.

[6] A. Manickam, A. Chevalier, M. McDermott, A. D. Ellington, and A.Hassibi, “A CMOS electrochemical impedance spectroscopy (EIS)biosensor array,” IEEE Trans. Biomed. Circuits Syst., vol. 4, no. 6, pp.379–390, 2010.

[7] Y. Mandel, S. Laufer, and B. Rubinsky, “Measurement of corneal en-dothelial impedance with non-invasive external electrodes a theoreticalstudy,” Med. Eng. Phys., vol. 34, no. 2, pp. 195–201, 2012.

[8] D. A. Schoeller, A. Alon, D. Manekas, L. A. Mixson, K. C. Lasseter,G. P. Noonan, J. A. Bolognese, S. B. Heymsfield, C. R. Beals, and I.Nunes, “Segmental bioimpedance for measuring amlodipine-inducedpedal edema: A placebo-controlled study,” Clin. Therapeut., vol. 34,no. 3, pp. 580–592.

[9] L. Hu, T. Maslanik, M. Zerebeckyj, and C. F. Plato, “Evaluation ofbioimpedance spectroscopy for the measurement of body fluid com-partment volumes in rats,” J. Pharm. Toxicolog. Methods, vol. 65, no.2, pp. 75–82, 2012.

[10] M. Steffen, A. Aleksandrowicz, and S. Leonhardt, “Mobile noncontactmonitoring of heart and lung activity,” IEEE Trans. Biomed. CircuitsSyst., vol. 1, no. 4, pp. 250–257, 2007.

[11] J. G. Webster, Medical Instrumentation: Application and Design, 4thed. Hoboken, NJ: Wiley, 2009.

[12] C. Gabriel, S. Gabriel, and E. Corthout, “The dielectric properties ofbiological tissues: I. Literature survey,” Phys. Med. Biol., vol. 41, pp.2231–2249, 1996a.

[13] S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties ofbiological tissues: II. Measurements in the frequency range 10 Hz to20 GHz,” Phys. Med. Biol., vol. 41, pp. 2251–69, 1996b.

[14] S. Grimnes and Ø. G. Martinsen, Bioimpedance and Bioelectricity Ba-sics, 2nd ed. San Diego, CA: Academic, 2008.

[15] V. Cherepenin, A. Karpov, A. Korjenevsky, V. Kornienko, A. Maza-letskaya, D. Mazourov, and J. Meister, “A 3D electrical impedance to-mography (EIT) system for breast cancer detection,” Phys. Meas., vol.22, pp. 9–18, 2001.

[16] R. Halter, A. Hartov, and K. D. Paulsen, “Design and implementationof a high frequency electrical impedance tomography system,” Phys.Meas., vol. 25, pp. 379–390, 2004.

[17] T. Kao, G. Boverman, B. S. Kim, D. Isaacson, G. J. Saulnier, J. C.Newell, M. H. Choi, R. H. Moore, and D. B. Kopans, “Regional ad-mittivity spectra with tomosynthesis images for breast cancer detec-tion: Preliminary patient study,” IEEE Trans. Med. Imag., vol. 27, pp.1762–1768, 2008.

[18] A. McEwan, J. Tapson, A. van Schaik, and D. S. Holder, “Code-di-vision-multiplexed electrical impedance tomography spectroscopy,”IEEE Trans. Biomed. Circuits Syst., vol. 3, no. 5, pp. 332–338, 2009.

[19] D. T. Ngyuen, C. Jin, A. Thiagalingam, and A. McEwan, “A review onelectrical impedance tomography for pulmonary perfusion imaging,”Phys. Meas., vol. 33, no. 5, pp. 695–706, 2012.

[20] J. Tehrani, C. Jin, and A. McEwan, “Modelling of an oesophageal elec-trode for cardiac function tomography,” Comput. Math. Methods Med.,2012, DOI: 10.1155/2012/585786.

[21] R.W.M. Smith, I. L. Freeston, B. H. Brown, and A.M. Sinton, “Designof a phase-sensitive detector to maximize signal-to-noise ratio in thepresence of Gaussian wideband noise,”Meas. Sci. Technol., vol. 3, pp.1054–1062, 1992.

[22] R. D. Cook, G. J. Saulnier, D. G. Gisser, J. G. Goble, J. C. Newell,and D. Isaacson, “ACT3: A high-speed, high-precision electricalimpedance tomography,” IEEE Trans. Biomed. Eng., vol. 41, pp.713–722, 1994.

[23] A. McEwan, G. Cusick, and D. S. Holder, “A review of errors in multi-frequency EIT instrumentation,” Phys. Meas., vol. 28, no. 7, 2007.

[24] B. J. McCarroll, C. G. Sodini, and H. S. Lee, “A high-speed CMOScomparator for use in an ADC,” IEEE J. Solid-State Circuits, vol. 23,no. 1, pp. 159–165, 1988.

[25] H. Y. Yang and R. Sarpeshkar, “A time-based energy-efficientanalog-to-digital converter,” IEEE J. Solid-State Circuits, vol. 40, no.8, pp. 1590–1601, 2005.

[26] L. Brooks and H. S. Lee, “A zero-crossing-based 8-bit 200 MS/spipelined ADC,” IEEE J Solid-State Circuits, vol. 42, no. 12, pp.2677–2687, 2007.

[27] C. S. Taillefer and G. W. Roberts, “Delta-sigma A/D conversion viatime-mode signal processing,” IEEE Trans. Circuits Syst. I, Reg. Pa-pers, vol. 56, no. 9, pp. 1908–1920, 2009.

[28] B. Razavi, Principles of Data Conversion System Design, 1st ed.Hoboken , NJ: Wiley, 1995.

[29] S. O’Driscoll, K. V. Shenoy, and T. H. Meng, “Adaptive resolutionADC array for an implantable neural sensor,” IEEE Trans. Biomed.Circuits Syst., vol. 5, no. 2, pp. 120–130, 2011.

[30] M. Yoshioka, K. Ishikawa, T. Takayama, and S. Tsukamoto, “A 10-b50-MS/s 820-W SAR ADC with on-chip digital calibration,” IEEETrans. Biomed. Circuits Syst., vol. 4, no. 6, pp. 410–416, 2010.

354 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 7, NO. 3, JUNE 2013

[31] S. Franco, Design With Operational Amplifiers and Analog IntegratedCircuits, 3rd ed. New York: McGraw-Hill, 2002.

[32] A. Stojcevski, H. P. Le, J. Singh, and A. Zayegh, “Flash ADC archi-tecture,” Electron Lett., vol. 39, no. 6, pp. 50–52, 2003.

[33] J. Daniels, W. Dehaene, M. S. J. Steyaert, and A. Wiesbauer, “A/Dconversion using asynchronous delta-sigma modulation and time-to-digital conversion,” IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 57,no. 9, pp. 2404–2412, 2010.

[34] Q. Liu, T. I. Oh, H. Wi, E. J. Lee, J. K. Seo, and E. J. Woo, “Designof a microscopic electrical impedance tomography system using twocurrent injections,” Phys. Meas., vol. 32, pp. 1505–1516, 2011.

[35] E. J. Lee, J. K. Seo, E. J. Woo, and T. Zhang, “Mathematical frameworkfor a new microscopic electrical impedance tomography system,” Inv.Prob., vol. 27, pp. 1–19, 2011.

Jong Cheol Baeg (S’12) received the B.S. andM.Eng. degrees in electrical engineering from theKorea Advanced Institute of Science and Technology(KAIST), Yuseong-gu, Daejeon, in 1992 and 1994,respectively.Since 2010, he has been working toward the

Ph.D degree at Kyung Hee University, Yongin-si,Gyeonggi-do, Korea. From 1994 to 1999, he workedfor Hyundai Electronics designing car electroniccontrol units. From 2000 to 2002, he worked withDSP applications for Texas Instruments Incorpo-

rated. Since 2002, he has been Chief Executive Officer of Syncworks Inc.,Yatab-dong, Bundang-gu, Seongnam-si, South Korea.

HunWi (S’12) received the B.S. andM.S. degrees inbiomedical engineering from Kyung Hee University,Yongin-si, Gyeonggi-do, Korea, in 2008 and 2010,respectively.Since 2010, he has been working toward the

Ph.D. degree at Kyung Hee University under thesupervision of Prof. Eung Je Woo. He is working onthe development of impedance imaging system. Hisresearch interests include analog and digital circuitdesign, biomedical instrumentation, electrodes,biomedical signal processing, and computing. He is

a member of KOSOMBE.

Tong In Oh (M’10) received the B.S. and M.S. de-grees in electronics engineering, and the Ph.D. degreein biomedical engineering from Kyung Hee Univer-sity, Yongin-si, Gyeonggi-do, Korea, in 1999, 2002,and 2006, respectively.From 2007 to 2009, he was a Post-Doctoral Re-

search Fellow in medical physics and engineering,University College London, London, U.K. Currently,he is an Assistant Professor in the Department ofBiomedical Engineering, Kyung Hee University. Hisresearch interests include analog and digital circuit

design, biomedical instrumentation, electrode and smart sensors, cell culturemonitoring, biomedical signal processing, and computing.Dr. Oh is amember of the IEEEEngineering inMedicine and Biology Society

and KOSOMBE.

Alistair Lee McEwan (M’89–SM’09) received theB.E., B.Com and M.Phil. degrees in economics andelectrical engineering in 1999 and 2001, respectively,from The University of Sydney, Sydney, Australia,and the D.Phil. degree in microelectronics from Ox-ford University, Oxford, U.K., in 2005.He is a Senior Lecturer of Computer Engineering

at The University of Sydney and InternationalScholar at Kyung Hee University, Yongin-si,Gyeonggi-do, Korea. He performed postdoctoralresearch at University College London, London,

U.K. and then at Philips Research Labs in Germany as a Marie Curie ResearchFellow. His research interests include medical instrumentation, integratedcircuit design, and bio-inspired systems. He has authored or coauthored morethan 80 papers in these research areas and is the inventor on nine patents.Dr. McEwan is a member of the BioCAS Technical Committee of the IEEE

CAS Society.

Eung Je Woo (M’83–SM’09) received the B.S. andM.S. degrees in electronics engineering from SeoulNational University, Seoul, Korea, and the Ph.D. de-gree in electrical and computer engineering from theUniversity of Wisconsin-Madison in 1983, 1985, and1990, respectively.From 1990 to 1999, he was an Assistant and

Associate Professor in the Department of Biomed-ical Engineering, Konkuk University, Seoul, Korea.Currently, he is a Professor in the Department ofBiomedical Engineering, Kyung Hee University,

Yongin-si, Gyeonggi-do, Korea. Since 2002, he has been the Director of theImpedance Imaging Research Center (IIRC). He teaches undergraduate andgraduate courses on bioinstrumentation, bioimaging, and inverse problems.His research interests include electromagnetic tissue property imaging of EIT,MREIT, MREPT and QSM, biomedical instrumentation, biomedical signalprocessing, and computing.Dr. Woo is a senior member of the IEEE Engineering in Medicine and Bi-

ology Society and a member of KOSOMBE.