ap-5-11.dvi Acta Polytechnica Vol. 51 No. 5/2011 Advances in Modern Capacitive ECG Systems for Continuous Cardiovascular Monitoring A. Schommartz, B. Eilebrecht, T. Wartzek, M. Walter, S. Leonhardt Abstract The technique of capacitive electrocardiography (cECG) is very promising in a flexible manner. Already integrated into several everydayobjects, the single lead cECGsystemhas shownthat easy-to-usemeasurements of electrocardiograms are possible without difficult preparation of the patients. Multi-channel cECG systems enable the extraction of ECG signals even in the presence of coupled interferences, due to the additional redundant information. Thus, this paper presents challenges for electronic hardware design to build on developments in recent years, going from the one-lead cECG system to multi-channel systems in order to provide robust measurements - e.g. even while driving an automobile. Keywords: capacitive electrodes, non-contactmeasurements, ECG,multi-channel sensor array, automotive application. 1 Introduction Cardiovascular diseases have for years been the most common cause of death (German Federal Statistical Office). Today, however, it is essential to monitor cardiac patients not only in hospitals, but also in ev- eryday life (e.g. while driving a car), because of the demographic change toward an aging population. In traffic situations, we do not risk only our own lives, and this makes monitoring even more important. Theelectrocardiogram(ECG)hasbeenknown for a long time as a fast examination tool which can pro- vide important clues to the status of the cardiovascu- lar system. It is often the tool of first choice in emer- gency situations, e.g. for detecting a heart attack. However, until now no extensive long-time monitor- ing system for people in high-risk groups could be developed, for technical and financial reasons. In recent years, a technique already known since 1967 through Richardson has been in the main re- search focus: measuringpotentialswith isolated elec- trodes [1]. This capacitive measurement method is nowadays built into a range of everyday objects: of- fice chairs [2,3], bathtubs [4], toilet seats [5], incuba- tors [6], cars [7,8]andbeds [9]. Asanexample, the so- called “AachenSmartChair” from [10] integrates two capacitive electrodes in an off-the-shelf office chair (see Figure 1) for easy-to-use measurements where no medical staff is needed and no difficult prepara- tion has to be done. Even a study on acceptability for medical staff and patients produced very positive results [11]. However, this technique has not yet been marketed, perhaps because most systems use a single lead, which is less robust thanconventional conductive sys- tems. Fig. 1: The “Aachen SmartChair”: A single lead cECG system in an office chair [10] This paper presents an overview of the capacitive measurement method, the challenges for electronic hardwaredesignandthedevelopments in recentyears fromsingle lead systems tomulti-channel systems for robust and reliable measurements. 2 Theory of capacitive ECG measurements Charge movements on the surface of human bodies, caused by their heart activity, influence the electric charge distribution on an electrode within a small distance from the body. This knowledge is used for capacitive measurements. Conductive electrodes, often made of Ag/AgCl, need a contact gel as a conductive contact to the 100 Acta Polytechnica Vol. 51 No. 5/2011 body surface to measure the potential between the two electrodes. This is mainly resistive behavior. With time, the contact gel dehydrates and reduces the quality of the signals. Capacitive electrodes, isolated e.g. by a very thin lacquer coatingwith high surface resistance, need no leading contact to the body. In addition, the po- tentials can be measured even through several layers of clothing (depending on material and thickness). Each electrode forms a coupling capacitance C with the patient’s body, which is known to be C = �0�r A d (1) where A is the effective surface area of the electrode, d is the thickness, �r is the dielectric constant of the clothes, and �0 is vacuum permittivity. With d = 0.3 mm (off-the-shelf cotton shirt), �r = 1 and A =40 mm×80 mm, as in [10], the coupling capaci- tance is about90pF.Forahigh couplingcapacitance, large electrode surfaces within small distances from the body are essential. One major disadvantage of this technique is the continuous changing of C due to patients’ move- ments, transpiration, varying thickness d and cloth- ingmaterials, aswell as static charges. In addition, it is a technical challenge to deal with high impedance biosignals in the rangeof a fewmV.Direct impedance conversion must be achieved in order to avoid dis- turbing voltages. 3 Realization of capacitive ECG The coupling capacitance forms the input of a high impedance input stage for impedance transfor- mation (Figure 2). Typically, a high-impedance bias resistor RBias (> GΩ) is connected to ground to dis- charge to capacitance, but this results in a first order highpassbehaviorwith the correspondingcut-off fre- quency: fc = 1 2πRBiasC . (2) Thus, variations of the coupling capacitances can re- sult in a shift of 0.1–100 Hz of the cut off frequency into the spectrum of the ECG, and can thus deterio- rate the ECG signal. Even the operational amplifier should have a very high input impedance for the same reason. Further on, due to the small voltage amplitudes, low noise amplifiers should be used and because of the shift of the operating point low bias currents are necessary. Signals measured at two different positions are combined, forming the differential input in an in- strumentationamplifierwithahighCMRR(common mode rejection ratio) of about 120 dB in order to fil- Fig. 2: Block diagram of the capacitive ECG system, modified from [10] Fig. 3: Matlab simulation of the resulting gain in the spectrum of the ECG after filtering ter the coupled power line interferences that underly both signals. Due to the asymmetries, e.g. variable lengthofwiresorvaryingdistancesof electrodes, cou- pled interference may have a slightly different effect. Therefore it cannot be eliminated in reality as well as in theory. Further filtering of the signal is indis- pensable. A low pass filter is used to cut off the high fre- quency components (typically > 100 Hz), which do not provide information for the ECG, as well as higher harmonics of the 50 Hz power baseline. With an additional notch filter at 50 Hz this noise can be suppressed. A high pass filter with a low cut off frequency of about 0.3 Hz deletes the DC compo- nents and minimizes the baseline drifts. Figure 3 shows the Matlab simulation of the filter and gain cascade, depending on the frequency, in the range of 0.1–1000 Hz. A nearly constant gain of 61 dB in the useful frequency range was achieved. The high pass filter with a cut off frequency of 0.3 Hz shows 101 Acta Polytechnica Vol. 51 No. 5/2011 moderateabsorption,whereas thehighpassof the ca- pacitive electrodes (due to uncertainties of the mod- elling depending on the distance between electrode and body) is not considered in this simulation. The notch filter at 50 Hz achieves suppression of 35 dB with regard to the desired signal. The remaining gain of 26 dB referring to the input signal is low enough. Higher suppression canbe achievedby addi- tional notch filter stages or digital signal processing. Controlling further components of the signal pro- cessing of the ECG signal, i.e. the A/D converter, significant amplification (about factor 1000) is usu- ally impliedatgoodSNR(signal tonoise ratio). First amplification can be realized by external circuitry of the instrumentational amplifier increasing the SNR without driving the subsequent components into sa- turation. The main amplification takes place at the end of the filter cascade in the output or gain stage. Figure 4 shows a realization of an ECG circuit with two simultaneous filter cascades. Fig. 4: Realization of anECGcircuit for two simultanous measurements of ECG Conventional ECG measurement systems make use of the principle knwon as “Driven Right Leg” to eliminate interference in the signal evenmore [12,13]. This principle can also achieve interference reduction in capacitive ECG measurements. There it is often referred to as “Driven Ground Plane”, as in [7], or “Driven Ground Electrode” (DGE), as in [14]. The negative feedback electrode connects the body and the output of an inverting operational amplifier the potential of which is the sumof the active electrodes, with amplificationof about −1000. In thisway, iden- tical potential changes on the body surface are not transferred to the system output (compare [7]). Dealing with these technical challenges alone will not ensure a robust cECG measurement system. Challenges due to the patient also need to be over- come. 4 Results Progress in digital circuit technology in recent years has enabled the implementation of complex algo- rithms in reasonable modules. Improved types of electrodes integrated in everyday objects have been achieved and there have been advances in signal pro- cessing. Fig. 5: Left: Experimental setup formulti-channel cECG, right: Multi-channel cECG integrated into the driver’s seat of a Ford automobile The latest cECG systems have been realized for multi-channel measurements. This redundancy may lead to robust measurements to ensure reliable med- ical statements on the basis of an ECG. As a first outcome, amulti-channel cECGsystemhasbeenpre- sented by [15], with 15 electrodes integrated in a tablet pc. However, the patient must additionally be connected to ground. To the best of the author’s knowledge, our grouppresented the first independent multi-channel cECG system with free choice for the negative feedback DGE to reduce the commonmode rejection [14]: on a square aluminum plate an ar- ray of nine round electrodes, positioned in a 3 × 3 matrix, was attached with flanged bearings enabling to tilt every electrode (Figure 5 left). In combination with the ability to have continuously adjustable posi- tions of the electrodeson theplate, this allowsproper adaptation to the silhouette of the patients. Ancil- lary connectors with springs have been developed to achieveproper adaptationand surfacepressure in ax- ial direction. This construction allows simultaneous measure- ments of up to eight cECGs where the leads can be chosen manually, just like the DGE. Figure 6 shows a sequence of six leads according to Einthoven (3) and Goldberger (4), measured on the patient’s back, which can be calculated as: I = U8 − U6 = φ8 − φ6, II = U8 = φ8 − φ1, III = U6 = φ6 − φ1, (3) aVR = U8 − U0 + U6 2 , aVL = U6 − U0 + U8 2 , aVF = U0 − U6 + U8 2 , (4) 102 Acta Polytechnica Vol. 51 No. 5/2011 where φi is the potential from the electrode i, φ1 is the reference potential and the voltage Ui = φi − φ1 (compare Figure 5 left). The sequence in Figure 6 shows that every channel has robust signals and the R-peaks can be clearly identified and therefore used e.g. for QRS detection. The possibility of developing a multi-channel cECG system enables integration into the driver’s seat (here a Ford S-Max) for cardiovascular moni- toring even in traffic situations (Figure 5 right). The positionsof the electrodes ina2×3matrix in theback rest were chosen due to pressure measurements with a flexible pressure sensor mattress by a group of ten males and femaleswithdifferentphysique. Averaging thepressuredistribution identified thebestpositions, and thesewere then verifiedby cECGmeasurements. A textile negative feedback electrode integrated into the seat panel is used for common mode reduction. Measurements of the pressure sensor mattress, with correspondingECGmeasurements, the best ca- pacitive channel and reference ECG as a gold stan- dard, for two persons, see Figure 7, show that a proper pressure contact to the electrodes is of es- sential interest. The BMI, a criterion for the rela- tion between weight and height of human beings, of the male proband in the upper part is normal with 22.2 kg/m2, and in the lower part the BMI is low with 16.8 kg/m2. It can be seen thatwith decreasing surface pressure, the signal quality of the cECG de- creased in comparisonwith the referenceECGbelow. Fig. 6: Sequenceof resulting leads according toEinthoven and Goldberger, as shown in (3), (4), transformed to the patient’s back Fig. 7: Pressure measurements of two subjects with different BMI: Male proband with normal BMI in the upper part, below female proband with low BMI. On the right side the cECG and the reference ECG as gold standard are shown for these persons 103 Acta Polytechnica Vol. 51 No. 5/2011 In test runs at aFord car test site, the systemwas validated in static tests andwhile driving ondifferent track surfaces. With 93%of the 59 probands, robust cECGs were measured in static tests. On motorway tracks, in particular, the detection rate of 92.4% is very robust [16]. City tracks, withmore steeringmo- tion and road damage, lead to lower detection rates than for the reference ECG. Importantly, irregulari- ties in theECGcould be detectedwithout restricting the driver’s level of comfort. 5 Conclusion In recent years, capacitive ECG measurement sys- tems developed by various groups have shown that this technique is very flexible due to integration into several everyday objects. However, the medical di- agnostics and the robustness of the systems are lim- ited becausemost systems are based on a single lead. Multi-channel cECGsystemspresent a promising ap- proach for monitoring people in high-risk groups. Even measurements while driving a car on a test track showed good performances. This is a major step towards monitoring and assisting drivers. Acknowledgement Parts of the research described in this paper have re- ceived funding from theEuropeanCommunity’s Sev- enth Framework Programme under grant agreement No. FP7-216695 and the Ford Forschungszentrum Aachen GmbH, Aachen, Germany. References [1] Richardson, P.: The insulated electrode. Pro- ceedings of the 20th Annual Conference on En- gineering in Medicine and Biology. Boston, 1967, vol. 157. [2] Lim,Y.,Kim,K.,Park,S.: ECGmeasurementon a chair without conductive contact. IEEE Trans- actions onBiomedical Engineering, 2006, vol.53, no. 5, p. 956–959. [3] Aleksandrowicz, A., Walter, M., Leonhardt, S.: Ein kabelfreies, kapazitiv gekoppeltes EKG- Messsystem/Wireless ECG measurement system with capacitive coupling. Biomedizinische Tech- nik, 2007, vol. 52, no. 2, p. 185–192. [4] Lim, Y., Kim, K., Park, K.: The ECG mea- surement in the bathtub using the insulated elec- trodes. 26th Annual International Conference of the IEEE EMBS, 2004, vol. 1. [5] Kim, K., Lim, Y., Park, K.: The electrically noncontacting ECG measurement on the toilet seatusing the capacitively-coupled insulated elec- trodes. 26th Annual International Conference of the IEEE EMBS, 2004, vol. 1. [6] Kato, T., Ueno, A., Kataoka, S., Hoshino, H., Ishiyama, Y.: An application of capacitive elec- trode for detecting electrocardiogramof neonates and infants. 28th Annual International Confer- ence of the IEEE EMBS, 2006, p. 916–919. [7] Leonhardt, S., Aleksandrowicz, A.: Non-contact ECG monitoring for automotive application. 5th International Summer School and Sympo- sium on Medical Devices and Biosensors, 2008, p. 183–185. [8] Chamadiya, B., Heuer, S., Hofmann, U., Wag- ner, M.: Towards a capacitively coupled electro- cardiography system for car seat integration. In Proceedings IFMBE, 2008, vol.22, p. 1217–1221. [9] Ishijima, M.: Monitoring of electrocardiograms in bed without utilizing body surface electrodes. IEEE Transactions on Biomedical Engineering, 1993, vol. 40, no. 6, p. 593–594. [10] Aleksandrowicz, A., Leonhardt, S.: Wireless and non-contact ECG measurement system — theAachenSmartChair.ActaPolytechnica, 2007, vol. 2, p. 68–71. [11] Czaplik, M., Eilebrecht, B., Ntouba, A., Wal- ter, M., Schauerte, P., Leonhardt, S., Ros- saint, R.: Clinical proof of practicability for an ECG device without any conductive con- tact, Biomedizinische Technik, 2010, vol. 55, p. 291–300. [12] Neuman, M.: Biopotential amplifiers, Medical Instrumentation: Application and Design, 1978, p. 292–296. [13] Winter, B., Webster, J.: Driven-right-leg circuit design, IEEE Transactions on Biomedical Engi- neering, 1983, p. 62–66. [14] Eilebrecht, B., Schommartz, A., Walter, M., Wartzek, T., Czaplik, M., Leonhardt, S.: A ca- pacitive ECG array with visual patient feed- back, 32nd Annual International Conference of the IEEE EMBS, 2010. [15] Oehler,M., Ling,V., Melhorn,K., Schilling,M.: A portable ECG system with capacitive sen- sors, Physiological Measurement, 2008, vol. 29, p. 783–793. [16] Eilebrecht, B., Wartzek, T., Lem, J., Vogt, R., Leonhardt, S.: Capacitive electrocardiogram measurement system in the driver seat, Auto- mobiltechnische Zeitschrift ATZ, 2011, vol. 113, p. 50–55. 104 Acta Polytechnica Vol. 51 No. 5/2011 About the authors Antje Schommartz was born in Essen, Germany, on March 31st, 1981. She studied Electrical Engi- neering and Information Technology, specializing in Medical Engineering, at Ruhr University Bochum, Germany, where she received her Dipl.-Ing. degree in 2009. She currently works as a Ph.D. student at thePhilipsChair ofMedical InformationTechnology, RWTH Aachen University, Germany. Her research interests are focused on capacitive ECG measure- ments and high frequency cardiac neuromodulation. She is a member of the VDE, DGBMT and IEEE. Benjamin Eilebrecht was born in Bochum, Ger- many, on June 6th, 1982. He studied Electrical En- gineering, with a specialization of Medical Engineer- ing, at Ruhr University Bochum, Germany, and re- ceived his Dipl.-Ing. degree in 2008. He is working as a Ph.D. candidate at the Philips Chair of Medi- cal Information Technology, RWTH Aachen Univer- sity, Aachen, Germany. His research interests in- cludenon-contactmonitoring techniques, learning al- gorithmsandmodeling. Hehasbeenamember of the German Electrical Engineering Association (VDE) since December 2008. Tobias Wartzekwas born inKrefeld in 1982. From 2003 to 2008, he studied electrical engineering fo- cusing on information and communication technol- ogy, and received his Dipl.-Ing. degree from RWTH Aachen University, Germany in 2008. He is cur- rently pursuing a Ph.D. degree at the Chair of Med- ical InformationTechnology, RWTHAachenUniver- sity, Aachen, Germany, where he is also working as a research assistant. His research interests are in the field of modeling physiological systems, new sensors for biomedical measurements, and automation and diagnosis support for the intensive care unit. MarianWalterwasborn inSaarbrücken,Germany, on March 4th, 1966. He studied Electrial Engineer- ing, with a specialization in Control Engineering, at the Technical University of Darmstadt and received his Dipl.-Ing. degree in 1995 and his Dr.-Ing. de- gree in 2002. He has worked in the medical engi- neering industry for three years and was appointed senior scientist and deputy head at the Philips Chair ofMedical InformationTechnologyatRWTHAachen University, Aachen, Germany, in 2004. His research interests include non-contact monitoring techniques, signal processing and feedback control in medicine. Steffen Leonhardt was born in Frankfurt, Ger- many, on Nov. 6th, 1961. He holds an M.S. in Com- puter Engineering from SUNY at Buffalo, NY, USA, a Dipl.-Ing. in Electrical Engineering and a Dr.- Ing. degree in Control Engineering from the Tech- nicalUniversity ofDarmstadt, Germany, and aM.D. in Medicine from J. W. Goethe University, Frank- furt, Germany. He has 5 years of R&D management experience in medical engineering industry and was appointed Full Professor and Head of the Philips en- dowed Chair of Medical Information Technology at RWTH Aachen University, Germany in 2003. His research interests include physiologicalmeasurement techniques, personal health care systems and feed- back control systems in medicine. Antje Schommartz E-mail: schommartz@hia.rwth-aachen.de Benjamin Eilebrecht Tobias Wartzek Marian Walter Steffen Leonhardt Philips Chair for Medical Information Technology RWTH Aachen University Pauwelsstr. 20, 52074 Aachen, Germany 105