DEFIBRILLATORS often use a small-signal ac measurement

Transcription

1 1858 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 61, NO. 6, JUNE 214 Bioimpedance-Based Respiration Monitoring With a Defibrillator Ørjan G. Martinsen, Senior Member, IEEE, Bernt Nordbotten, Sverre Grimnes, Helge Fossan, and Joar Eilevstjønn, Member, IEEE Abstract Defibrillators often use an ac measurement to ensure safe electrode placement. Some defibrillators also utilize impedance measurements for monitoring. This paper investigates how such measurements can be optimized for high respiration sensitivity and finds that it is beneficial to add two extra electrodes in addition to the two defibrillator electrodes. This four electrode setup increases sensitivity and also allows respiration measurements at lower frequencies than the two electrode setup. Index Terms Bioimpedance, defibrillator, respiration, sensitivity,. I. INTRODUCTION DEFIBRILLATORS often use a small-signal ac measurement current in the range khz in order to ensure good electrode placement. In this frequency range, the impedance between defibrillator pads on a human body is predominantly resistive and normally matches the impedance opposing the defibrillator current quite well. Typical impedances of an Apex-Sternum placement of defibrillator pads are in the 25 2 Ω range, and defibrillation is normally restricted to an impedance range close to this. Some defibrillators have implemented extra functionality based on the measured impedance; most common is monitoring which is done by measuring the variation of the base impedance caused by. The modulation is normally below ±3 Ω and the signal can be obscured by motion artifacts. It is believed that detection/monitoring can be improved if the impedance measurement system was more optimized for this task. It is the purpose of this study to find whether the respirational plethysmographic signal can be increased by using a Manuscript received December, 213; revised February 9, 214; accepted February 2, 214. Date of publication February 28, 214; date of current version May 15, 214. Asterisk indicates corresponding author. Ø. G. Martinsen is with the Department of Physics, University of Oslo, Oslo 316, Norway, and also with the Department of Clinical and Biomedical Engineering, Oslo University Hospital, Oslo 424, Norway ( ogm@fys.uio.no). B. Nordbotten is with the Department of Physics, University of Oslo, Oslo 316, Norway ( berntn@gmail.com). S. Grimnes is with the Department of Physics, University of Oslo, Oslo 316, Norway, and also with the Department of Clinical and Biomedical Engineering, Oslo University Hospital, Oslo 424, Norway ( sverre.grimnes@fys.uio.no) H. Fossan and J. Eilevstjønn are with Laerdal Medical, Stavanger 42, Norway ( helge.fossan@laerdal.no; joar.eilevstjonn@laerdal.no). Digital Object Identifier.19/TBME four-electrode system [1], and to compare the admittance and impedance parameters. II. MATERIALS AND METHODS Ten healthy male volunteers participated in this study. Average age was 3 years (21 4). They all gave informed consent and the study was approved by the regional ethics committee. All measurements were conducted in a controlled laboratory with 22 C and 3% RH. A. Impedance Measurements All impedance measurements were done on healthy male volunteers, using a Solartron 126/1294 combination with a controlled ac of 3 mv rms. At maximum inspiration or maximum expiration, respectively, one complex spectrum took about 45 s during which the test held his breath. (Note that this time could have been reduced, e.g., by utilizing time domain measurements with multisine or chirp signal excitation [2].) B. Defibrillator Measurements A Philips HeartStart MRx, which has a controlled ac of 25 μa rms at 32 khz for impedance measurement, was used for the defibrillator measurements. The MRx is an advanced monitor/defibrillator device with 12 lead ECG, blood pressure, ETCO2, pulse-oximetry, networking capability, and CPR feedback (QCPR) functionality. C. Electrode Positioning The Apex-Sternum electrode placement is shown in Fig. 1. The defibrillator electrodes Philips Medical DP2/DP6 and the small ECG electrodes are Ambu Blue Sensor Q A. The defibrillator electrodes were only used for the defibrillator measurements and for the Solatron two-electrode setup. The Blue Sensor electrodes were used as voltage pick-up electrodes in the four-electrode setup. The edge-to-edge distance between current injecting and voltage pick-up electrodes is about 1 cm. D. Forced Breathing and Generation of Motion Artifacts For the time-series measurements, it was measured both during spontaneous breathing where the test subjects breathed freely, and also during forced breathing where the subjects respiration was controlled with an even rate. For the forced breathing a manually operated calibration syringe with a volume of 75 ml, together with a valve system which allowed the IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See standards/publications/rights/index.html for more information.

2 MARTINSEN et al.: BIOIMPEDANCE-BASED RESPIRATION MONITORING WITH A DEFIBRILLATOR 1859 TABLE I DEFIBRILLATOR MEASUREMENT RESULTS Fig. 1. Electrode placement for respiration monitoring. Large defibrillator electrodes used in the two-electrode setup and as current injecting electrodes in the four-electrode setup. Small ECG electrodes used as potential pick-up electrodes in the four-electrode setup. inhalation to be controlled while allowing for free exhalation, was used. The pump was fitted with a single use mouth piece for each test subject. Motion artifacts can be due to compression-induced changes in either the electrode materials or in the measured tissue [3], and different electrodes and electrode systems may be more or less prone to producing motion artifacts [4]. In this study, motion artifacts were generated by pushing on the electrodes; either by hand on the lower electrode or using the MRx CPR sensor as an interface between the hand and upper electrode. E. Protocol The following protocol was followed for all the ten test subjects: 1) The sat comfortably in a chair with no clothing on the upper body. After a few minutes of acclimatization, the electrodes were positioned. 2) The defibrillator was then connected to the two defibrillator electrodes and impedance measurements were done continuously through the following sequence: 3 s of spontaneous breathing, 6 s of forced breathing, a few seconds of spontaneous breathing, and then 6 s of forced breathing while repeatedly pressing on one of the defibrillator electrodes. 3) The was then instead connected to the Solartron 126/1294 system in a two-electrode setup, and the above sequence was repeated, using a frequency of 32 khz. The was then asked to breathe maximally in and keep his breath during a frequency scan from Hz to 1 MHz. Subsequently, the breathed maximally out and kept his breath while a new frequency scan was conducted. 4) Lastly the same procedure, including the frequency scans, was repeated with a four-electrode setup % % % % % % % % % % % TABLE II TWO-ELECTRODE RESISTANCE TIME SERIES RESULTS % % % % % % % % % % % III. RESULTS A. Defibrillator Measurements The results from the measured time-series with the defibrillator can be seen in Table I. The table shows the total number of s, the mean of all measured values in the time series (mean ), the mean of the impedance change for all s (mean ) with standard deviation, and the relative. The mean relative was found to be 1.8%. B. Solatron 126/1294 Measurements For the time series measurements, only the 6-s time period with forced breathing was since during this period the test subject has a fairly even and reproducible respiration rate. 1) Two-Electrode Setup: The results from the two-electrode time series measurements are shown in Table II. Only resistance values are. The mean relative was found to be 2.5%. A frequency scan with the two-electrode setup is shown as mean values in Fig. 2, both at maximum inspiration and expiration. The relative difference between maximum expiration and inspiration is shown in Figs. 3 and 4 as percent change in impedance parameters (resistance and reactance) and admittance parameters (conductance and susceptance), respectively.

3 186 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 61, NO. 6, JUNE 214 TABLE III FOUR-ELECTRODE RESISTANCE TIME SERIES RESULTS Fig. 2. Impedance modulus and phase spectra (mean values) for maximum inspiration measured with the two-electrode set-up shown in Fig. 1. Dashed lines are maximum expiration values. Error bars show standard deviation. Impedance % % % % % % % % % % % Impedance Phase angle Phase angle [deg.] Fig. 3. Two-electrode impedance measurements. Percent impedance increase from maximum expiration to maximum inspiration held breath values. Dashed line indicates extreme values due to measurement uncertainty/error. Admi ance increase [%] Conductance Susceptance 1 1 Fig. 4. Percent admittance increase (or decrease when negative values) from maximum expiration to maximum inspiration held breath values for twoelectrode measurements. Dashed line indicates extreme values due to measurement uncertainty/error. Fig. 5. Transfer impedance, mean values. Modulus and phase spectra measured with the four-electrode method. Continuous lines are maximum inspiration values and dashed lines are maximum expiration values. Error bars show standard deviation. C. Four-Electrode Setup The results from the four-electrode time series measurements are shown in Table III. Only resistance values are. The mean relative was found to be 4.%. The results from the frequency scans using a four-electrode system [1] are shown as mean values in Figs Data from one of the ten test subjects have been removed, since the lowfrequency reactance values for maximum expiration were more than ten times higher than for any of the other subjects, and the corresponding resistance about four times higher, hence indicating loose electrodes. 1) Noise Measurements: Fig. 8 shows a typical time course of the measured resistance using the two-electrode system. The last part of the time series shows noise caused by repeatedly pressing one of the defibrillator electrodes. By visual inspection of the curves, no significant difference in the noise contribution between the two- and four-electrode measurements was found. IV. DISCUSSION Figs. 2 and 3 support the choice of 32 khz for the impedance measurements in the defibrillator. However, from the difference

4 MARTINSEN et al.: BIOIMPEDANCE-BASED RESPIRATION MONITORING WITH A DEFIBRILLATOR 1861 Impedance increase [%] Resistance Reactance 1 1 Fig. 6. Percent transfer impedance increase from maximum expiration to maximum inspiration held breath values. Four-electrode method. Dashed line indicates extreme values due to measurement uncertainty/error. Fig. 7. Percent transfer admittance increase from maximum expiration to maximum inspiration held breath values. Four-electrode method. Dashed line indicates extreme values due to measurement uncertainty/error. Fig. 8. Typical time course of resistance at 32 khz measured with twoelectrode setup. in the mean relative between the twoand four-electrode setup using the Solartron 126/1294 system, there is an indication that the four-electrode setup is better suited for picking up the small resistance changes due to respiration. The differences between the two- and four-electrode setup will now be further discussed. In the two-electrode setup, the reactance increase diverges to a very high value (359%) at 46 khz. This is due to a very small phase angle as the phase is actually crossing zero in that frequency range. At this frequency, the data did not meet the normality criterion and a one-sample Sign for the distribution of relative reactance increase at 46 khz revealed that the median increase was not significantly different from zero (p = 1.). Hence, what appears as very high sensitivity for the reactance and susceptance at high frequencies is actually an artifact due to small (random noise) changes in very small numbers. The two-electrode results are clearly dominated by skin impedance (23 Ω at Hz) [5] in the frequency range Hz, Fig. 2, which of course is not very dependent on respiration, Figs. 3 and 4. At higher frequencies, the results are determined by deeper tissue layers and the sensitivity to respiration increases. The highest sensitivity can be attained by measuring resistance (or equally well impedance since the phase angle is very small) at frequencies around khz or higher where a relative increase of around 18% can be achieved, or alternatively by measuring susceptance in the frequency range 5 khz where the decrease is 2%. The same problem with low-phase angle as described with the two-electrode system is present for low frequencies with the four-electrode system. Extremely low ( 122%) and high (57%) average relative increases were found in the reactance at the two lowest frequencies, respectively. At these frequencies, the data did not meet the normality criterion and a one-sample Sign for the distribution of relative reactance increase at and 215 Hz revealed that the median increase was not significantly different from zero (p =.51 for both). In fact the reactance increase was not significantly different from zero for any frequency below khz. For the susceptance, the extreme value at Hz ( 1%) is not significantly different from zero (p =.18) whereas values at 215 Hz (2%) and 464 Hz ( 21%) are marginally different from zero (p =.4). The four-electrode method measures transfer impedance [6], resulting in very low values around 4 Ω, which are clearly not influenced by skin impedance. An appreciable respiration sensitivity of around 18% (see Figs. 6 and 7) is therefore possible even at Hz. As for the two-electrode measurements, it is also found here that resistance or impedance measurements at high frequencies (above khz) yield good sensitivity (3% 35%). The highest sensitivity, however, is found for susceptance measurements between 2 and khz, where a relative increase of 35% was achieved. For the four-electrode measurements, using the impedance modulus yields basically the same results as using resistance, since the phase angle is very small. For the two-electrode measurements, the results for using only the impedance modulus give a mean relative that is about 1% lower than for the resistance throughout the frequency range. Only male subjects were measured in this study. This may possibly influence the conclusions about the sensitivity of the monitoring of the respiration, since the thoracic impedance is influenced by the female chest [7]. No significant difference in the noise contribution between the two- and four-electrode measurements was found. However, it should be noted that a four-electrode system in some cases

5 1862 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 61, NO. 6, JUNE 214 may produce more noise than a two-electrode system, e.g., in the case of drying voltage pick-up electrodes in combination with a very high input impedance differential amplifier. V. CONCLUSION It has been shown that the addition of two extra electrodes is beneficial for the measurement of respiration. With the two electrode setup, the increase in resistance due to respiration is only 1.38% at Hz and increases to 18.18% at 1 MHz, while for the four electrode setup the increase in resistance is 18.4% at Hz and increases to 34.16% at 1 MHz. It is furthermore concluded that with the four-electrode setup, it is also possible to measure the respiration at low frequencies, whereas for the two-electrode setup the measurements at low frequencies are dominated by the skin impedance. No significant difference was found between using impedance modulus and resistance values, but a small improvement of sensitivity could be achieved by using the susceptance. [2] J. Ojarand and M. Min, Simple and efficient excitation signals for fast impedance spectroscopy, Electron. Elect. Eng.,vol.19,pp.49 52,213. [3] B. Belmont, R. E. Dodde, and A. J. Shih, Impedance of tissue-mimicking phantom material under compression, J. Electr. Bioimp.,vol.4,pp.2 12, 213. [4] S. Luo, V. X. Afonso, J. G. Webster, and W. J. Tompkins, The electrode system in impedance-based measurement, IEEE Trans. Biomed. Eng., vol. 39, no. 11, pp , Nov [5] Ø. G. Martinsen, S. Grimnes, and E. Haug, Measuring depth depends on frequency in electrical skin impedance measurements, Skin Res. Technol., vol. 5, pp , [6] S. Grimnes and Ø. G. Martinsen, Sources of error in tetrapolar impedance measurements on biomaterials and other ionic conductors, J. Phys. D: Appl. Phys., vol. 4, pp. 9 14, 27. [7] J. H. Meijer, J. P. H. Reulen, P. L. Oe, W. Allon, L. G. Thijs, and H. Schneider, Differential impedance plethysmography for measuring thoracic impedances, Med. Biol. Eng. Comput., vol. 2, pp , REFERENCES [1] H. P. Schwan, Determination of biological impedances, in Physical Techniques in Biological Research, W. L. Nastuk, Ed. New York, NY, USA: Academic Press, 1963, vol. 6, pp Authors photographs and biographies not available at the time of publication.