Wireless Brain Signal Recordings based on ...

2 downloads 70 Views 997KB Size Report
Index Terms—Brain signal, electroencephalogram monitoring system, capacitive .... 5 shows the top layer of the capacitive electrode board from Eagle layout ...
Wireless Brain Signal Recordings based on Capacitive Electrodes Mehrnaz Kh. Hazrati * Graduate School for Computing in Medicine and Life Sciences, Institute for Signal Processing University of Lübeck Lübeck, Germany [email protected]

Haliza Mat Husin

Ulrich G. Hofmann

Institute for Signal Processing University of Lübeck Lübeck, Germany

Neuroelectronic Systems, Department of Neurosurgery, University Medical Center Freiburg Freiburg, Germany [email protected]

Abstract— In this paper the development of a wireless electroencephalogram (EEG) monitoring system is presented. The system is capable of processing brain signals on-board recorded from non-contact sensors. The non-contact sensor was designed utilizing capacitive coupling as recording interface. The on-board multi-channel signal processing is performed on a tiny computer module with low power consumption, high performance embedded computing platform that can communicate via WiFi or Bluetooth. It provides an excellent option for developing a compact Brain Computer Interface (BCI) with a direct connection to the external device e. g. robot or prosthesis without employing a personal computer (PC).

Index Terms—Brain signal, electroencephalogram monitoring system, capacitive electrode, Overo Fire, Gumstix®.

I. INTRODUCTION Electroencephalography (EEG) is one of the most widely used methods for evaluating the electrical activities of the brain. Due to the advantages of non-invasive measurement and the capability of long term monitoring of the EEG signal, an electroencephalograph plays an important role in brain examinations and studies. EEG is a standard procedure used in clinical and research applications, especially, in the diagnosis of brain diseases such as epilepsy, sleeping disorder and abnormal behavior [1]. EEG measurements can be divided into two types based on the manner in which electric current is collected from the tissue and potential differences are measured; (1) contact sensors (also called electrodes) are typically made of metal and need to come in direct electrical contact with the skin [10, 11]. This galvanic contact implies that there is a transfer of electric current between the sensors and the tissue. In that case electric currents are transferred through an electricallyconductive gel. Gelled elctrodes are often referred to as wet selectrodes, otherwise they are called dry electrodes [1, 11], (2) capacitive sensors are equipped with an insulated conducting plate placed in close contact with the body’s surface. This plate and the body’s surface form a capacitor and the electric signal can be derived over it. This is referred to as capacitive-coupling measurement [2, 5].

Two issues limit the use of commercially available EEG recording systems. The primary limiting issue is that they are hardly portable. Commercially available EEG recording systems are usually wired to a stationary computer, in order to save and to analyze the data. In addition recording electrodes are usually mounted on a tight cap that the subject wears on the head. The EEG cap is not portable due to the mass of wires connecting the cap to the data collecting system, which itself has a high power consumption. The second problem is that electrically-conductive gel is often required for a good connection between the mounted sensors and the scalp. The gel takes a lot of time to apply and it tends to dry out, which limits the recording time. Dry electrodes depend on ionic contact at the electrode-electrolyte interface as for the wet electrode type. The difference is that the skin supplies the electrolyte in the form of perspiration [2], rather than relying on an artificial electrolyte. The impedance of dry electrodes is much higher than that for wet electrodes, so a buffer must be used to convert high impedance to low impedance in order to minimize the effects of the various noise sources, which are the main issue in designing these electrodes. As an alternative, capacitively coupled electrodes have an advantage over galvanic contact electrodes: they can be used without electrolyte gel on the unprepared skin and are immune to voltage drifts that could appear because the electrode-skin resistance changes [2, 4, 8]. To tackle the mentioned issues, one approach is the use of capacitive electrodes to provide analog brain signal sensing [2] along with a small lightweight embedded computer for signal processing on the head. In this work, we introduce a wireless EEG monitoring system which is capable of processing brain signals on-board recorded from non-contact sensors. In the following the elements of the system are introduced and some recorded data are validated. II.DESIGN OF CAPACITIVE ELECTRODES Capacitive electrodes allow acquiring bio-potentials through displacement currents instead of real charge currents since the electrolyte-electrode-skin interface is replaced by a dielectric insulating film. Due to the absence of electrolyte, capacitive electrodes have a different behavior concerning the skin contact. A capacitive method of picking up biopotentials was proposed in

1967 by Richardson [2] where the active electrode concept was taken into account. The results demonstrated that capacitive electrodes can be used to pick-up ECG signals with good signal characteristics in comparison to wet electrodes. Up till now, capacitive electrodes have mostly being studied for ECG recordings [7, 13, 16, 18]. The new procedure based on capacitive measurements of skull potentials, while no direct electrical contact with the scalp is made and therefore no gel is needed, was first reported by Matsuo et al in 1973 using barium titanate as the insulating material [3]. In 2004, recordings of ECG signals on a toilet seat have been performed by Kim et al. at the Advanced Biometric Research Center, Seoul National University [19]. Their capacitively-coupled electrode is composed of a Cu plate and a PTFE film. Kim et al. (2008) also made an important contribution [12] by introducing a capacitively-coupled active ground using and extending the driven-right-leg scheme described by Webster in 1983 [5]. In particular, they showed that an active ground is highly effective at reducing line noise. Researchers at Quantum Applied Science and Research (QUASAR) developed a sensor that is able to measure the ECG of a fully clothed person standing within a range of about 25 cm. In 2005, QUASAR developed a compact version of the sensor and named it the capacitively coupled noncontact electrode (CCNE), specifically to measure ECG through clothing [16]. Sullivan et al. (2007) from Institute for Neural Computation, University of California San Diego, designed an integrated sensor which combines amplifications, band-pass filtering, and analog-to-digital conversion within a 1 inch diameter enclosure [9]. This non-contact bio-potential sensor couples capacitively to the human scalp through hair for EEG and to chest through clothing for ECG recordings [13, 18]. A solid copper fill forms a parallel plate capacitor with the body and works as a capacitive electrode by sensing signals through insulation such as fabric. In 2008 at the “Institut für Elektrische Messtechnik und Grundlagen der Elektrotechnik” of the TU Braunschweig, Oehler et al. designed a capacitive electrode for EEG measurements through hair [7]. 28 electrodes were integrated into an adjustable helmet to allow for different head shapes. The state of the art of capacitive electrodes was reviewed in the work of Spinelli and Haberman in 2010 [14]. They designed capacitive electrodes that allow detection of biopotentials through thin clothes. On the other hand, they reported that a sophisticated shielding and guarding at the front-end stages is required to reduce high movement artifacts and power line interferences. Recently Chi et al. designed an innovative micro power non-contact EEG electrode with active common-mode noise suppression and input capacitance cancellation [17]. Our active electrode concept has adopted the capacitive method by Richardson [2]. In contrast to the previous built electrodes, we employed a combination of low noise, low power electronic components such as an instrumentation amplifier LTC6079 and capacitance cancellation scheme.

III. SYSTEM DESIGN The proposed EEG monitoring system is composed of two main components: an analog unit running multi-channel capacitive electrodes which include actively driven grounding, and a digital unit based on a tiny computer module, Overo Fire made by Gumstix® that has the capability to communicate via WiFi and Bluetooth [22]. The system block diagram is shown in Fig. 1 acquiring scalp potentials with the analog unit which in turn is connected to the digital unit responsible for analog to digital conversion and signal processing.

Fig. 1. Block diagram of the wireless EEG recording system using Gumstix commercial computer unit and Summit extension board

A. Analog Unit The capacitive electrode was designed so that its input impedance would be significantly larger than that of the skinelectrode impedance to minimize interference caused by motion artifact and unwanted common-mode voltages. The signal on the skin capacitively couples to the sensing plate coated with dielectric material to achieve the capacitive effect. The coupling capacitance depends mainly on the thickness and the dielectric constant of the material located between the electrode and the subject’s skin [4]. The electrode surface forms a coupling capacitance between the subject’s body and the electrode as shown in Fig.2. The coupling capacitance depends mainly on the thickness and the dielectric constant, of the dielectric material located between the electrode and the subject’s skin. Using a capacitor model formula, the amount of capacitive coupling, present can be estimated, where is the relative static permittivity of dielectric, is the permittivity of free space, is the surface area of the plates and is the thickness of dielectric.

Fig. 2. Capacitive sensing method

For our electrode, the signal on the skin capacitively couples to the sensing plate at the bottom of the PC board, which is covered with soldering mask for electrical isolation of the sensor. The effective surface area of the 30 mm diameter electrode board is A = π.r2 = 709 mm2. To achieve the capacitive effect, the sensing plate had to be coated with dielectric. For simplicity, a cellophane tape based on cellulose was chosen as dielectric with a dielectric constant of 3.9 [23] and the thickness of 0.058 mm. By using the capacitor model equation, the approximate amount of capacitive coupling that could be achieve by an electrode with diameter 30 mm is:

(a)

The general structure of a capacitive electrode is simplified as in Fig. 3. The implementing circuit design behind a capacitive electrode neccessarily needs to achieve low current noise levels [4, 6]. Fig. 4 shows the top of our finished electrode design and the connection between two units. The electronic circuit that is attached to the backside of the sensing plate of the electrode incorporates a complex combination of circuit techniques in order to achieve the required low current noise levels [5, 6]. It was built based on operational amplifiers with very high input impedance in the order of TΩ to PΩ so the lower frequency limit of EEG signals is not compromised [5]. The dominance of the resistive contribution to the total input impedance requires an input biasing circuit to maintain a stable dc operating point [5, 6]. Neutralization of input capacitance is needed to ensure the electrode’s gain is constant over a wide range of coupling distances thus making the device applicable for diagnostic use [4]. Active shielding also known as guarding, shields the electrode and electronics to prevent the influence of disruptive external electric signals [7].

Fig. 3. General structure of capacitive electrode

The PCB design for the capacitive electrode consists of a 30 mm diameter four-layer PCB. The top layer houses the component and traces, second layer is ground plate, third layer is shielding plate and the bottom layer is the sensor plate. Such PCB stacks are recommended for capacitive coupled sensing applications [39, 40]. Fig. 4(a) presents each PCB layer of the electrode, whereas Fig. 4 (b) illustrates the thickness of copper and isolation for each of these layers.

(b) Fig. 4. (a) Four-layer PCB of capacitive electrodes, layer 1 is for traces, layer 2 is the ground plane, layer 3 is the shielding plate and layer 4 is the sensing plate, (b) Copper and Isolation thickness for PCB layers in Eagle design software.

Number 1 and 16 represent the top and bottom layer respectively and number 2 and 15 are the 2 layers in between. Fig. 5 shows the top layer of the capacitive electrode board from Eagle layout and the actual board. The reference DRL electrode was also constructed of a four-layer PCB but with with 19 mm in diameter smaller in size compared to capacitive electrodes. This is due to the less complex circuit required for this electrode. To have a fully capacitive measurement system its bottom plate connects to the body capacitively like the . other electrodes. B. Digital Unit For the digital unit, we chose to develop on the Overo Fire, a Linux-based computer-on-module (COM) produced by Gumstix®. The Overo is a fully functional computer motherboard that uses the Texas Instruments OMAP 3503 Application Processor. With roughly the size of a stick of gum, this tiny board includes DSP (TMS320C64x) and GPU (ARM Cortex-A8) as well as Bluetooth and the 802.11g wireless networking protocols [21, 22]. The Overo Fire is an appropriate embedded computing platform within our system‘s requirements because of its minimal power requirements. As consequence of its minute size and weight it can be placed on patient’s head and still has an expansion port to add sensors. The Summit is an expansion board that helps to support the features of the Overo. It features a power supply for both itself and the Overo [22]. The connection with the analog sensor array is established via the ADC pin of the Summit board. The 10-bit ADC lines at the Overo board are controlled by the TPS65950 Audio and Power Management module with a maximum input voltage of 2.5V [21, 22].

(a)

D. The Connector Board The sensor connector board bridges the Summit and Overo Fire board with sensors, and was also responsible for supplying power to sensors. With the dimension of 80 mm x 39 mm, it is exactly the same size as the Summit board and allows the two boards to be be stacked together thus providing extra mechanical stability. EEG data was stored into an external SD-memory card on the Gumstix. Finally the sensor data was manually transferred to the host machine. Connectivity between sensors and Gumstix was provided by a flexible flat cable through a custom board connected with a 40-pin header to the Summit board where the ADC pins are located. The ADC driver at the Overo kernel was interrupted using a system call made at the Linux userspace. Fig.6 shows the PCB of the connector board. The top layer of the board is the signal routing of analog power supply, sensor input and output, shielding and analog ground node for all electrodes. The bottom layer is where all the low contact connectors coming from electrodes are placed.

(b) Fig. 5. (a)Top view of the built capacitive electrode, (b) Analog and digital units.

Since the Overo is a development platform, the appropriate kernels to drive the analog sensors had to be configured, next to programming the acquisition algorithms. C. Gumstix ADC Analog to Digital Converter (ADC) lines run direct from the TPS65950 Power Management module in the Overo Fire board with an input range of 0 to 2.5 V. ADC’s pin 2-7 can be accessed at the 40-pin header (SV1) at the Summit expansion board [35,36]. The TPS65950 platform device supports numerous functions which most of them are power related [9, 11]. It also supports the ADC module through twl4030-madc driver and is connected to the OMAP chip on the I2C bus. TWL3040 is a comparable older Texas Instruments chip which uses uses the same ADC module as the TPS65950 [35]. ADC driver can be accessed through a device input output control (ioctl) interface, a single system call by which userspace can communicate directly with the device driver [35, 37]. This ioctl function was used to send control codes to the ADC. There are sample programs provided by the developer to access ADC driver depending on the version of Linux kernel run on the Gumstix [22].

Fig. 6. The sensor connector board bridges the Summit and Overo Fire board with sensors, and was also responsible for supplying power to sensors

II. RESULTS Before designing the hardware, an AC sweep simulation was applied in PSpice to evaluate the skin-electrode-analog unit based on an existing skin-electrode model. The simulated sensor circuit provides differential gain over a bandwidth of 1100Hz. In practice, any 50 Hz voltages from the power supply line may cause a common mode voltage on the subject’s body. Capacitive, actively driven ground connection using a wellknown technique by Webster [5] was applied for the purpose of reducing the power line noise. We further applied a notch filter at 50 Hz to minimize the effect of this noise in recorded signals. Fig. 7 shows a time-domain signal of 8 seconds of EEG data collected from Ag-AgCl and capacitive electrode at Fp1 during an eyes closed, relax period. The red line represents the EEG signal from capacitive electrodes, while the blue line represents the EEG signal from the classical conductive electrode. The bottom window in each figure represents a plot of one second signal extracted from the same EEG data. Correlation coefficient between two signals during eye closed state was calculated 0.7169 and this value was 0.6853 during eye open state.

technique in acquiring very low amplitude of brain signal eliminates the need for electrolyte gel and skin abrasion. Future work of the sensor will include the mechanical stability of the electrode placement on the body to improve signal quality as the next step towards a portable monitoring device. Overall, we consider the combination of custom made analog and off-the-shelve embedded system to provide the basis for continuous brain monitoring for extended periods of time. The final device will feature a compact design; low power consumption as well as efficient data transmission and processing due to its on-chip DSP and the option to directly connect any sensor to the Overo. It is an excellent option for developing a Brain Computer Interface (BCI) with a direct connection to the external device e. g. robot or prosthesis without employing a personal computer (PC). Fig. 7. A comparison between designed capacitive electrode and a commercial electrode: Recorded EEG of Fp2 while subject closed his eyes, eight seconds samples (top) and one second samples (bottom).

Fig. 8 illustrates a 2min recorded signal from the occipital position, when the subject closed his eyes for 60 sec in the middle of the recording. Note the increase in both amplitude and α activity (around 8-10 Hz as seen in the zoom of the spectrogram Fig. 6).

30 20 10 0

30

40

50

Fig. 8. Time series and spectrogram of 30-60-30 eye activities recorded with the capacitive electrode. α activity can served be around 8-10 Hz.

III. CONCLUSION An EEG monitoring system for non-contact sensing has been presented in this paper. The use of capacitive coupling

REFERENCES Niedemeyer and F.H Lopes Da Silva, [1] E. Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. Lippincott Williams & Wilkins, 2005. [2] A. Lopez and P. C. Richardson, “Capacitive electrocardiographic and bioelectric electrodes,” IEEE Transactions on Biomedical Engineering, vol. BME-16, no.1, pp. 99–99, Jan. 1969. [3] T. Matsuo, K. Iinuma, and M. Esashi, “A barium-titanateceramics capacitive-type EEG electrode,” IEEE Trans. Biomed. Eng., vol. BME-20, no. 4, pp. 299–300, Jul. 1973. [4] K. Larry and K. Baxter, Capacitive Sensors: Design and Applications. John Wiley and Sons, 1996. [5] B. B. Winter and J. G. Webster, “Driven-right-leg circuit design,” IEEE Transactions on Biomedical Engineering, vol. BME-30, no. 1, pp. 62–66, Jan. 1983. [6] C. J. Harland, T. D. Clark, and R. J. Prance, “Electric potential probes - New directions in the remote sensing of the human body”, Measurement Science and Technology, vol. 13, no. 2, p. 163, 2002. [7] M. Oehler, V. Ling, K. Melhorn, and M. Schilling, “A multichannel portable ECG system with capacitive sensors,” Physiological Measurement, vol. 29, no. 7, p. 783, 2008. [8] Y. M. Chi and G. Cauwenberghs, G. et al “Dry-Contact and Noncontact Biopotential Electrodes: Methodological Review,” IEEE Reviews in Biomedical Engineering, vol. 3. 2010. [9] T. Sullivan, S. Deiss, and G. Cauwenberghs, “A low-noise, noncontact EEG/ECG sensor,” in Proc. IEEE Biomedical Circuits Systems Conf., pp. 154–157, 2007 [10] A. Searle and L. Kirkup, “A direct comparison of wet, dry and insulating bioelectric recording electrodes,” Physiological Measure., vol. 21, no. 2, p. 271, 2000. [11] D. Prutchi and M. Norris, “Design and Development of Medical Electronic Instrumentation: A Practical Perspective of the Design, Construction and test of Medical Devices”, John Wiley & Son, 2005. [12] K. K. Kim and K. S. Park, “Effective coupling impedance for power line interference in capacitive-coupled ECG measurement system,” in Proc. Int. Conf. Information Technology Applic. Biomedicine ITAB, pp. 256–258, 2008. [13] Chamadiya, B., K. Mankodiya, M. Wagner and U. G. Hofmann "Textile-based, contactless ECG monitoring for non-ICU

[14]

[15]

[16]

[17]

[18]

clinical settings." Journal of Ambient Intelligence and Humanized Computing, 2012 E. Spinell and M, Haberman, “Insulating electrodes: a review on biopotential front ends for dielectric skin-electrode interfaces” in Physiological Measure., pp 183-198, 2010. R. Mattews, N. J. McDonald, I. Fridman, P. Hervieux and T Nielsen, “The invisible electrode – zero prep time, ultra low capacitive sensing” in International Conference on Human Interaction, 2005. Y. M. Chi and G. Cauwenberghs, “Wireless non-contact EEG/ECG electrodes for body sensor networks,” in Proc. Int. Conf. Body Sensor Networks (BSN), pp. 297–301, 2010. Y. M. Chi and G. Cauwenberghs.“Micropower non-contact EEG electrode with active common-mode noise suppression and input capacitance cancellation.” in Engineering in Medicine and Biology Society, 2009.EMBC 2009. Annual International Conference of the IEEE, pages 4218 –4221, 3-6 2009. Y. M. Chi, S. Deiss, and G. Cauwenberghs, “Non-contact low power EEG/ECG electrode for high density wearable biopotential sensor networks,” in Proc. 6th Int. Workshop

Wearable Implantable Body Sensor Networks BSN, 3–5, pp. 246–250, 2009. [19] K. K. Kim, Y. K. Lim, and K. S. Park, “The electrically noncontact in ECG measurement on the toilet seat using the capacitively-coupled insulated electrodes,” in Proc. 26th Annu. Int. Conf. IEEE Eng. Medicine Biol. Soc., vol. 1, pp. 2375– 2378, 2004. [20] TPS65950 Integrated Power Management/Audio Codec, Datasheet: Texas Instrument, April 2008 [21] OMAP 35xx Technical Reference Manual: Texas Instrument, April 2010. [22] Gumstix,www.gumstix. com [23] Delta Controls Corporation, “Dielectric Constants of Various Material”, Application Notes

Suggest Documents