IOP PUBLISHING
PHYSIOLOGICAL MEASUREMENT
Physiol. Meas. 34 (2013) N1–N7
doi:10.1088/0967-3334/34/1/N1
NOTE
Towards true unipolar bio-potential recording: a preliminary result for ECG Gaetano D Gargiulo 1,2,3 , Alistair L McEwan 2 , Paolo Bifulco 3 , Mario Cesarelli 3 , Craig Jin 2 , Jonathan Tapson 1 , Aravinda Thiagalingam 2 and Andr´e van Schaik 1 1 2 3
The MARCS Institute, The University of Western Sydney, Sydney, NSW, Australia EIE School, The University of Sydney, Sydney, NSW, Australia DIBET ‘Federico II’ The University of Naples, Naples, Italy
E-mail:
[email protected]
Received 13 September 2012, accepted for publication 31 October 2012 Published 18 December 2012 Online at stacks.iop.org/PM/34/N1 Abstract We present a bio-potential front-end capable of recording unipolar ECG leads without making use of the Wilson central terminal (WCT). The information contained in the new unipolar recordings may yield unique diagnostic information as it avoids the need to essentially subtract data or make use of the averaging effect imposed by the WCT. The system also allows a direct, real-time software calculation of signals corresponding to standard ECG leads for standard diagnosis. These calculated standard ECG leads have a correlation in excess of 92% with a gold standard ECG recorded in parallel. The circuit is wideband, compatible with both the standard and the dry electrodes, and of low power (requiring less than 20 mW powered at 12 V). It is therefore well suited for long-term applications. Keywords: ECG, bio-potential recording, unipolar bio-potential (Some figures may appear in colour only in the online journal) 1. Introduction Standard bio-potential amplifiers (such as for ECG) rely on a differential electrode configuration which consists of two sensing electrodes placed across the subject’s body to form a sensing lead. Often a third electrode, the driven right leg (DRL) electrode, is used as a reference for common-mode voltage reduction. Therefore, any ECG lead can be regarded as the result of a double-difference operation. First, each component (such as left and right arms composing Lead I) is a result of the voltage difference between one point and the reference; second, a differential amplifier calculates the difference between the two components. Traditionally, the use of a differential amplifier is regarded as an advantage because of the high input impedance and high common mode 0967-3334/13/010001+07$33.00
© 2013 Institute of Physics and Engineering in Medicine
Printed in the UK & the USA
N1
N2
Note
rejection ratio (CMRR). However, differences in the signal path, such as contact impedance variation, can cause a fraction of the large common mode signal to be treated as a differential signal and hence can be amplified (Webster 2006, Enderle 2006, Webster 1998). In the same way, a small differential mode signal may appear as a common voltage on both electrodes and hence can be erased (Bronzino 2000). Traditional differential ECG recordings are often degraded by the presence of very low frequency noise. This low frequency noise is composed of several noise components such as electrodic noise, thermal noise, electrode polarization and the contact impedance imbalance. Of these components probably the contact impedance imbalance is the greatest one (Webster 1998). Since bio-potential measurements involve very small and finite current flow in the measuring circuit (thru the electrode), the reason why contact impedance, intended as skin–electrode interface impedance, plays such a crucial role in electrophysiology (as general recommendation its measured value has to be lower than 5 k) is understandable. Unfortunately, despite the numerous recommendations to contain its value, such as skin cleaning, skin abrasion and use of conductive gel, it is inevitable that there will be an imbalance between the measured contact impedance of different electrodes; moreover, gel desiccation, skin stretching, movements and sweat can also contribute in a random fluctuation of the contact impedance imbalance which is recommended to be kept within the value of 1 k (Webster 2006, Enderle 2006, Webster 1998). The interposition of a high-pass filtering stage between the electrodes and the inputs of the amplifier, although useful in containing such noise, is not recommended as it would degrade the input impedance (Gargiulo et al 2011). Normally, the gain of the pre-amplifier stage is kept very low and filtering is implemented in further stages (Gargiulo et al 2011). Additional precautions to further reduce the noise capture and to increase the total CMRR of the bio-amplifier have been developed during the years. Some of the most important and widely used precautions are the following: active guard shields for cables, DRL and voltage supply bootstrap (Horowitz and Hill 2002, Gargiulo et al 2011). Particularly, the latter can be used to create a virtual ground potential (referred to the common mode noise) which may make the grounding connection to the subject redundant (Gargiulo et al 2011, Webster 2006). In traditional electrocardiography, the so-called unipolar leads (precordial) make use of the Wilson central terminal (WCT) as a reference. However, since the WCT is derived from the centre of the limbs’ star connection (the three electrodes connected to the limbs are tied together by supposedly identical high-value resistors), not just it gives an overall average body potential but since the limbs are also used as ECG leads may also contain some ECG information. Since the WCT is derived by multiple electrodes which require being independently prepared, it can also be a source of noise and signal artefacts due to the already mentioned contact impedance imbalance between the limbs electrodes and the precordial ones (Webster 1998, Degen and Jackel 2008, Bronzino 2000). We present a different unipolar approach to ECG recording which does not require the second potential difference, the high-pass noise filtering at the inputs; this does not make use of the WCT as a reference point and is immune from the contact impedance imbalance by design.
2. Methods In figure 1, we show the bio-potential amplifier functional block diagram. We regard the local bio-potential measurement as a combined observation of noise and useful signal and that it
Note
N3
Figure 1. Front-end functional block diagram.
is thus possible to measure the local signal of interest by subtracting the local noise from the measured signal. While this operation may appear to be trivial, one can note that the huge common mode signal affecting physiological measurements, since it is not measured at two different locations over the subject’s body, has to be regarded as an absolute measurement. Hence, its cancellation has to be operated from the combined action of a carefully designed noise filter (see figure 1) and a low-noise instrumentation amplifier corroborated by other noise-reduction techniques such as careful cable shielding (Horowitz and Hill 2002), active electrode shielding (Gargiulo et al 2008, 2010) and voltage supply bootstrap (Winter and Webster 1983). 2.1. Background knowledge For this design, in particular, in order to avoid additional noise capture by the capacitive coupling of disturbance signals on the electrode leads, active guard (driven shielded cable) has been implemented as recommended in the literature driving the cable shield with a buffered copy of the signal detected at the electrode (see figure 1) namely minimizing the capacitive noise coupling (Horowitz and Hill 2002). The driven shielding has also been extended to the electrode stud to protect against the capacitive noise coupling also the skin area under the active plate of the electrode (Gargiulo et al 2010, 2011). Although the majority of the bio-potential amplifiers make use of the DRL as an additional measure to reduce the noise (Bronzino 2000, Webster 1998), in this design, we made the grounding connection redundant maintaining the option to connect it in the case of a very noisy environment (see figure 1). Ground connection suppression is normally achieved by implementing the voltage bootstrapping technique (Winter and Webster 1983). In our design, we implemented an improved version of this technique (Gargiulo et al 2011, 2010). Namely, the voltage bootstrap in this case is to be intended as a ‘forced variation’ of the reference voltage and is imposed by the average potential of all the sensing electrodes. Since the differential elements (Horowitz and Hill 2002, Sedra and Smith 2004) are referred to this reference value, one way to increase the CMRR is to feed a common pattern or a signal to the reference terminal. In our circuit, the average measured common mode signal is fed to the voltage reference input of the differential amplifier via a damping RC filter, which reduces saturation events (Webster 1998, Gargiulo et al 2011). The average common mode signal contains the
N4
Note 1
LEAD I LA RA
0.8
Amplitude (mV)
0.6 0.4 0.2 0 -0.2 -0.4 -0.6 2.3
2.4
2.5
2.6 2.7 Time (Seconds)
2.8
2.9
3
Figure 2. ECG recorded by our device from a patient simulator Medsim 300B (settings: 1 mVpp @ 80 bpm); solid grey trace: right arm; dashed grey trace: left arm; solid black trace: computercomputed Lead I (RA–LA).
1 LA RA LL Computed lead I
R 0.8 Extra masked peak
Amplitude (mV)
0.6
T
0.4 P
0.2 0
S Q
-0.2 -0.4 -0.6
0
0.1
0.2
0.3 0.4 Time (Seconds)
0.5
0.6
0.7
Figure 3. Human recordings, visual inspection of LA (solid black line), RA (dashed black line), LL (solid grey line) recorded unipolarly and computed Lead I (solid red line).
potentially large baseline variation caused by contact impedance imbalance and electrodic noise at each electrode. 2.2. Novel contribution As is possible to observe from figure 1, the faint signal mixed with the noise is buffered, protected against noise pickup using the mentioned active guard circuitry (Horowitz and Hill 2002) and fed into a carefully designed filter stage. Necessary signal return is provided via a voltage supply bootstrap, which as mentioned, may make the grounding connection to the
Note
N5 1.5
Lead I RA LA Computed Lead I
Amplitude (mV)
1
0.5
0
-0.5
0.1
0.2
0.3
0.4
0.5
0.6
Time (Seconds)
Panel (a) 0.6 Lead II LL Computed Lead III RA
0.4
Amplitude (mV)
0.2 0 -0.2 -0.4 -0.6 -0.8
0.1
0.2
0.3
0.4
0.5
0.6
Time (Seconds)
Panel (b) 1 Computed Lead III LL Lead II LA
0.8
Amplitude (mV)
0.6 0.4 0.2 0 -0.2 -0.4 -0.6 0.1
0.2
0.3
0.4
0.5
0.6
Time (Seconds)
Panel (c) Figure 4. Human recordings: direct comparison of calculated standard ECG, blue line Leads I (panel a), Lead II (panel b) and Lead III (panel c) with traditional ‘golden standard’ ECG leads recorded in parallel, red lines.
body redundant. Note that the grounding circuitry (consequentially the grounding electrode) can be shared between several channels in multi-electrodes montages (Gargiulo et al 2011). Given that most of the unwanted noise in ECG recording is in the very low frequency (i.e. contact impedance variation, skin stretching, gel desiccation), we designed the noise filter as
N6
Note 0.6 0.4
Amplitude (mV)
0.2 0 -0.2 -0.4
V1 V1 unipolar
-0.6 -0.8 -1 0.1
0.2
0.3 0.4 Time (Seconds)
0.5
0.6
Figure 5. Human recordings: direct comparison of standard ECG V1 (black line), and unipolar V1 (grey line) showing evident distortion due to the use of the WCT as a reference (i.e. the R-peak is delayed of several ms).
low pass (active single order 0.005 Hz cut-off frequency) and fed its output to the inverting input of the instrumentation amplifier. This approach has the sought high-pass effect on the output signal. Because in our design the noise filter is only a first order, it does not introduce phase distortion in the transition bandwidth (see figure 1). To ensure subject safety, a calibrated coupling impedance is interposed between the electrode connection and the amplifier inputs. This impedance protects the circuit at the same time against electrostatic discharges (e.g. at connection with the skin) and the subject from micro-shock ensuring that even in the worstcase scenario (power supply shorted with the amplifier inputs) a current not larger than 100 μA can flow through the electrodes (Gargiulo et al 2011, 2010, Webster 1998). Besides the traditional bench testing, we evaluated our circuit with a parallel recording against a gold standard differential system (limited to limbs, leads and the precordial lead V1) using skin surface electrodes. 3. Results The point-to-point correlation between the sample signal generated by the ECG simulator and the calculated leads was 0.9. A data excerpt representing the calculation of Lead I from the left arm (LA) and right arm (RA) components is represented in figure 2. The point-to-point correlation has been calculated between the calculated Lead I (black bold line in figure 2) and the ‘high ECG level output’ of the Medsim 300b which make available an amplified version of the simulated Lead I. An example of the unipolar limbs leads compared with the computed Lead I is shown in figure 3. As is possible to infer from the figure, traditional differential ECG can mask features (like some extra peaks only visible from the RA recording) that may open some new diagnostic opportunities. A direct comparison of the three computed limbs leads with the parallel recorded traditional leads showed a point-to-point correlation of 0.95. A data excerpt showing the direct comparison of the calculated limb leads against the golden standard recordings is depicted in figure 4. A data excerpt showing a direct comparison of the precordial lead V1 with the unipolar signal recorded by nearby electrodes is depicted in figure 5. As is possible to observe from the
Note
N7
figure, the traditional differential signal is affected from distortions resulting in an enhancement of the P wave visibility and a delay in the occurrence of the R-peak. 4. Conclusion We have presented the preliminary evaluation of a novel bio-potential front-end circuit that does not rely on the conventional differential signal between two points across the subject’s body, referred to a supposedly neutral point, to detect a bio-potential. Moreover, recording bio-potentials by components (unipolarly) makes the proposed front-end immune from the contact impedance imbalance by design. The quality of the recording achievable using the proposed front-end and the proposed new local noise cancellation technique has been shown to agree with a gold standard ECG in two sets of tests. The diagnostic advantages (if any) still need to be investigated. In particular, concerning the appearance, possibly new diagnostic features like the extra masked peak visible in the QRS complex depicted in figure 3. Acknowledgment We gratefully acknowledge Texas Instruments that kindly provided us samples for our prototype development. References Bronzino J D (ed) 2000 The Biomedical Engineering Hand Book 2nd edn (Boca Raton, FL: CRC Press) Degen T and Jackel H 2008 Continuous monitoring of electrode–skin impedance mismatch during bioelectric recordings IEEE Trans. Biomed. Eng. 55 5 Enderle J D 2006 Bioinstrumentation (San Mateo, CA: Morgan and Claypool) Gargiulo G, Bifulco P, Calvo R A, Cesarelli M, Jin C and Schaik A V 2008 A mobile EEG system with dry electrodes Proc. IEEE Biomedical Circuits and Systems Conf. (Baltimore, MD, USA, 20–22 Nov.) pp 273–76 Gargiulo G, Bifulco P, Calvo R A, Cesarelli M, Mcewan A, Jin C, Ruffo M, Romano M, Shephard R and Schaik A V 2011 Giga-Ohm high-impedance FET input amplifiers for dry electrode biosensor circuits and systems Integrated Microsystems: Electronics, Photonics, and Biotechnology ed K Iniewski (Boca Raton, FL: CRC Press) Gargiulo G, Calvo R A, Bifulco P, Cesarelli M, Jin C, Mohamed A and Schaik A V 2010 A new EEG recording system for passive dry electrodes Clin. Neurophysiol. 121 Horowitz P and Hill W 2002 The Art of Electronics (Cambridge: Cambridge University Press) Sedra A S and Smith K C 2004 Microelectronic Circuits (Oxford: Oxford University Press) Webster J G (ed) 1998 Medical Instrumentation Application and Design (New York: Wiley) Webster J G (ed) 2006 Encyclopedia of Medical Devices and Instrumentation vol 3 (New York: Wiley) Winter B B and Webster J G 1983 Reduction of interference due to common mode voltage in biopotential amplifiers IEEE Trans. Biomed. Eng. BME-30 58–62
