HSI 2009

.

Catania, Italy, May 21-23, 2009

Towards a Wearable Device for Deep Brain Signals Monitoring F. Amarù†, P. Arena†, A. Latteri†, D. Lombardo†§, P. Mazzone‡, G. Vagliasindi†, †DIEES – Università di Catania, v.le A. Doria 6, Catania, Italy, §STMicroelectronics Castelletto, 20010 Cornaredo (MI), Italy, ‡C.T.O. “A. Alesini” via S. Nemesio, 21 - 00145 Roma, Italy

Abstract — Deep brain stimulation (DBS) is a surgical treatment for movement disorders like Parkison’s disease or essential tremor. Although DBS is effective in the treatment, the mechanism of action is not clearly understood by now. The accessibility to deep brain regions through the DBS implanted electrodes can give the opportunity to monitor the local activity of nuclei once accessible only during surgical operations. Monitoring extensively the signals from these regions can constitute an added value in understanding the mechanism of action of movement disorders as well as DBS. In this work a first prototype of wearable device for DBS recordings is proposed. The proposed device is portable and modular and demonstrated to be suitable in conditioning extremely low voltage signals like local field potentials (LFP) are. Keywords — Wearable biomedical devices, deep brain stimulation, local field potentials.

I

I. INTRODUCTION

N neurotechnology, deep brain stimulation (DBS) is a surgical treatment used as therapy in movement disorders such as Parkinson’s disease (PD) and essential tremor (ET), especially for patients with advanced symptoms. Also dystonia and tremor associated with cerebellar lesions seem to achieve benefits from stereotactic surgery. According to the designed anatomic target or the surgical methodology used, we can distinguish different categories of stereotactic surgery. Potential targets are the ventral intermediate nucleus (VIM) of the thalamus for tremor and the internal globus pallidus (GPI) and the subthalamic nucleus (STN) for PD, while GPI is the preferred target for dystonia. [1] The advantage of using DBS when compared to other surgical techniques like direct lesioning consists firstly in its reversible nature. Moreover the stimulus parameters can be adjusted as often as needed to achieve maximal benefit. At the same time, potential serious side effects such as dysarthria or aphonia can be reduced with DBS because the stimulus parameters on one or both sides can be adjusted or, if needed, one side can be entirely turned off. Contact authors are Guido Vagliasindi ([email protected]) and Alberta Latteri ([email protected]) , Dipartimento di Ingegneria Elettrica, Elettronica e dei Sistemi, v.le A. Doria 6, Edificio 3, 95125, Catania, Italy.

The implantation of electrodes gave the opportunity to access brain regions previously accessible only during surgical operation. This has opened new perspectives in the studying of Parkinson disease. The acquisition of the signals through the DBS electrodes permits, indeed, to analyse and study the activity of the implanted nuclei and provides new data for the assessment of the mechanism of action of both the disease and the stimulation parameter and protocol. Acquire those signals in a suitable way is, then, essential in order to make appropriate conclusions from their analysis. Although there are already several commercial devices devoted to biosignals acquisition, they are usually bulky [2] or just transportable [3], allowing the acquisition only in ambulatories and with limited patient’s motor capabilities since he is obliged to stay close to the instrument. A miniaturization of the signal conditioning and recording instrument, instead, could allow the acquisition of LFP signals also during the normal patient’s activity. The data so retrieved could be exploited to derive a correlation between the exogenous symptoms and deep brain signals, useful to derive a model of basal ganglia dyamics. A first step towards a portable device capable of conditioning the LFP signals and recording them into a flash memory is here proposed. Since it is a first prototype it was designed pointing the attention on the modularity, in order to test various circuit implementation. In section II and III the Modular MEDICAID is described in all its components while section IV and V report some of the tests performed on the device. Finally the conclusions are drawn. II.

THE HARDWARE SYSTEM

Medical data acquisition systems, in particular the patient monitoring systems, present the challenge of measuring very small electrical signals in presence of much larger common mode voltages and noise [4]. This is particularly true when considering neuron extracellular potentials. The signal is constituted by spike with amplitude about 40-250 μV, with a signal-to-noise ratio equal to 2-20 being the noise level typically below 10 μV RMS. The frequency range of signals comprises frequencies from few Hz up to 2 kHz. For the internal neuronal activity, the values of amplitude may be few μV,

978-1-4244-3960-7/09/ $25.00 2009 IEEE

Authorized licensed use limited to: UNIVERSITA DI CATANIA. Downloaded on August 18, 2009 at 05:50 from IEEE Xplore. Restrictions apply.

128

and the . frequency components of basal ganglia are below 100Hz [5]. Therefore, it is necessary to ensure a high gain, a good frequency response and a good rejection of disturbances, since the biological signal amplitude is very small, while at the same time the system should be safe both for the user and the patient. The proposed hardware implementation, the Modular MEDICAID, tries to meet all these requirements maintaining, at the same time, a limited dimension.

• • • •

Right Leg Driver Circuit; Notch Filter and Variable Gain Amplifier; Hybrid Low Pass Filter; Butterworth Low Pass Filter

III. MODULAR MEDICAID The Modular MEDICAID is an evolution of a previous version developed in DIEES laboratories [6]. The main improvement introduced is the modularity. The Modular MEDICAID is, indeed, constituted by a main board, hosting the power supply circuitry and the input and output connectors, where all the “modules” can be integrated through custom connectors. A picture of the board is shown in Fig. 1.

Fig. 2. Schematic representation of the Modular MEDICAID.

In the following a brief description of all the modules is proposed. When not otherwise stated, the active components used are low cost precision operational amplifiers TS507 (100uV as maximum offset voltage and 12nV/Hz at 1kHz as noise) from STMicroelectronics. A. ESD Protection Circuit For medical devices, protection against high voltage is fundamental. When the electrodes are connected to a patient, hazardous voltages and currents may arise from static discharges [8], which can destroy the high input impedance amplifiers and damage the system. A good performance commercial product for this purpose is the ESDA6V1-4BC6 [9] from STMicroelectronics. It is a component which provides an high ESD protection level maintaining at the same time very limited dimensions.

Fig 1. Picture of the Modular MEDICAID. The board dimensions are 10.5 cm by 8.5 cm.

The power supply circuitry requires a supply voltage between 8 V and 12 V and provides a regulated output voltage of 5V, required by all the components in the modules. The input voltage can be provided by batteries or external power supplies. The power supply stage is electrically insulated from the remainder part of the board using an isolated DC/DC converter (TSM0505S) from Traco Power [7]. The operating point of the conditioning circuit was centered at 1.63 V for a dual purpose: on one side to work with only single supply components, on the other to provide an output signal comprised between 0 and 3.3 V in order to easily interface it with an analog to digital converter. The modules of the Modular MEDICAID are the following: • ESD (Electro-Static Discharges) Protection Circuit; • Instrumentation Amplifier;

B. Instrumentation Amplifier The objective of this circuit is to adapt the input impedance of the measurement terminals (sensing electrode) to the subsequent stage and, also, to suppress common mode voltages. The filtered differential biological signal from the protection circuit is sent to an integrated amplifier for instrumentation, the INA118 from Texas Instruments. It has an extremely accurate, stable and optimized voltage gain, so as to operate properly in hostile environments to measure, even in the case of particularly unfavourable signal-to-noise ratios [10]. The input-output relation is described by equation (1): ⎛ 1 + 50 KOhm ⎞ ⎟(Vin+ − Vin − ) Vout = ⎜ ⎜ ⎟ Rg ⎝ ⎠

(1)

where Vout is the output voltage, (Vin+-Vin-) is the differential input voltage and Rg is a gain resistor. Varying the resistor Rg, it is possible to increase the differential gain without amplifying the common mode of input signal. The actual configuration provide a differential static gain of 21.7 dB. C. Right Leg Drive Circuit To further reduce the common mode voltage, a Right Leg Drive (RLD) circuit was used [11]. This circuit reduces common mode voltage by using a negative feedback loop. The common mode voltage coming from 129

Authorized licensed use limited to: UNIVERSITA DI CATANIA. Downloaded on August 18, 2009 at 05:50 from IEEE Xplore. Restrictions apply.

E. Hybrid Low Pass Filter This board is a low-pass filters in hybrid technology with fc = 59Hz. It is a third order hybrid filter as it integrates the main features of the Bessel and Butterworth filter [14]. The hybrid filter, designed in this board, has a more flat group delay than a Butterworth, but not maximally flat as that of Bessel filter. However, its frequency response has, at the border frequency between pass-band and transition band, a ripple more marked than the Butterworth, however, steeper than the Bessel filter. The introduction of this filter, moreover, increases the gain of 24 dB. Afterwards the hybrid filter a further pole was introduced, simply consisting of a pair capacitor-resistor, which increases the order of the low pass filter. From Bode plots can be noticed that the slope of roll-off is equal to 19.2 dB/octave, which corresponds to a 3.2 order filter. F. Butterworth Low Pass Filter Another low-pass filter, alternative to the previous one, was also foreseen. It is a fourth order Butterworth filter implemented with two Sallen-Key cells [14] in noninverting configuration providing a gain equal to 2.3 dB. It has a linear response in the frequency domain with a cutoff frequency of 96.23 Hz. Figure 3 show the frequency response of the board with different configurations: the Butterworth low-pass filter with and without notch and the Hybrid low-pass filter with and without notch. The selectable gain was fixed to 61.7 dB. It is possible to observe the effect of the passive notch filter which attenuates the frequencies around 50 Hz. IV. SAFETY TEST Although the system is just a prototype, it was decided

V. IN VITRO AND IN VIVO ACQUISITIONS In order to assess the performance of the device in terms of amplification capability and reliability, in vitro and in vivo acquisitions were performed. A. In Vitro The in vitro experimental setup consists in a waveform generator providing a sinusoidal signal with an amplitude of 500 μV peak to peak and a frequency of 10 Hz. The signal was provided as differential input to the Modular MEDICAID and its output was then acquired using a National Instruments USB acquisition board [17] interfaced with LabView. Several configurations of the Modular MEDICAID were evaluated, varying both the variable gain on the amplifier and the type of low-pass filter (Butterworth, Hybrid). The device response was in every case congruent with the expected behaviour. 90 80 Magnitude [dB]

D. Notch Filter and Rotary Switch Variable Gain Amplifier A notch filter centred at 50Hz and an amplifier with a variable gain constitutes this board. One of the problems in the acquisition and processing of biomedical signals is the electromagnetic interference from power ac line at 50 Hz. Unfortunately, this frequency is inside the useful band of biomedical signals so it cannot be cut-off with a simple low-pass or high-pass filter. The solution to remove unwanted frequency is given by the band-stop or notch filter. A simple implementation of notch filter is the twin-T passive filter [12], which requires only three resistors and three capacitors. The adoption of the notch filter can be enabled/disabled through a jumper. An operational amplifier in non-inverting configuration represents the second block of the circuit. Several configurations were designed which can be selected through a rotary switch to provide different amplification gains. The minimum selectable gain is 21.7 dB, while the maximum achievable gain is 81.7 dB.

to perform the safety test, to assess the accomplishment of the proposed instrument to norms related to medical electrical equipment. According to IEC norm 60601-1 (third edition), approved by CENELEC as EN 60601-1, related to general requirements for basic safety and essential performance of medical electrical equipment, the maximum patient leakage current is 10 μA [15]. The safety test consists in measuring the current that flows on the Signal Input Part (SIP), in this case the electrodes, during the acquisition of signals. The current was measured with a precision multimeter Keithley 2602 [16], which can measure currents ranging from 1pA to 3.06A, connected in series with the acquisition electrodes. The leakage current in the electrodes was measured in two conditions: when providing a test sinusoidal signal from a waveform generator and when acquiring an Electrocardiogram (ECG) from a volunteer. In the first case the input signal was a sinusoid with an amplitude of 1 mV peak to peak and the measured leakage current was around 0.6 μA. In the second case, the patient leakage current is in the order of few pA, well below the norm prescriptions.

70 60 50 40 30 20

Phase [deg]

the electrodes is inverted and amplified back into the body . though a third electrode in order to reduce the electrodeskin impedance

10 -1 10 200 100 0 -100 -200 -300 -400 -500 -600 -700 -1 10

Butterworth+Notch ButterWorth Hybrid+Notch Hybrid 0

10

0

10

1

10

1

10 Frequency [Hz]

2

10

2

10

3

10

3

10

Fig. 3. Bode plot of the board with different low pass filter configuration (Butterworth, Hybrid), with and without the notch filter and with the variable gain at 61.7 dB.

130

Authorized licensed use limited to: UNIVERSITA DI CATANIA. Downloaded on August 18, 2009 at 05:50 from IEEE Xplore. Restrictions apply.

B. In . Vivo The subsequent validation step was the acquisition of a biological signal from living beings. To perform this step we exploited the facilities available at the laboratories of the Department of Physiological Sciences of the University of Catania. We used one of the laboratory cavy rats available to acquire an electrocardiogram. Although the device was mainly designed for LFP acquisition, its modular nature permits to easily modify the main parameters in order to adapt it to a wide range, in terms of frequency and amplitude, of signals. At the current stage, for the present unavailability of the instrumentation necessary to reach the Basal Ganglia position in the rat brain, and since the rat was already being used for other experiments, we decided to acquire an electrocardiogram. Although the signal characteristics are different from LPF, nevertheless they are well known, so it can be easily understood whether the device is properly working or not. The ECG was acquired using an intra-epidermal needle electrode and during the whole experiment the rat was anaesthetized to minimize its movements.

extensive post-operative acquisition of LFP from deep brain. The device was designed to be adaptable and wearable. Thanks to its modularity, the device can be easily adapted to other biomedical signals like electrocardiographic or electroencephalographic ones. The portability of the device is granted owing to its limited dimensions (10.5 by 8.5 cm) that could be further reduced effortlessly, since the actual main board exploits just the top layer. Moreover the board autonomy is granted by the capability to be powered by standard 9V batteries. The output of the device was designed to be in the 0-3.3 V range in order to be easily converted in digital format. The portability of the proposed hardware, indeed, could be further increased integrating an analog to digital converter and a local storage device or a wireless module. In this way the presence of an external acquisition board and storage device, at the moment constituted by a NI USB board and a laptop, physically connected to the Modular MEDICAID output would be unnecessary. The data would be stored locally on a flash drive to be analyzed subsequently or sent through a wireless connection (WiFi, Bluetooth) to a remote server where they could be stored and analyzed. ACKNOWLEDGMENT The authors would like to acknowledge Prof. Guido Li Volsi, and his group for the kind support and hospitality during in vivo experiments. REFERENCES [1]

Fig. 4. ECG from the laboratory rat. In red is the typical electrocardiogram trend with the P, Q, R. S and T waveform highlighted. A frequency sample rate of 1000 samples/s was used.

Figure 4 shows a portion of the whole acquisition. In red is the typical P wave, QRS complex and T wave that characterize a normal ECG. It can be noticed that the dynamic of the acquired signal follows clearly the PQRST heart beat waveforms. To further validate our device, the acquisition was also performed with the WPI ISO-80 biomedical amplifier [3], a commercially available device. Its outputs suitably matched those ones of Modular MEDICAID. It is worth pointing out that the monitoring instrumentation applied to the cavy during signal acquisition did not reveal any damage that could have been potentially caused by the application of the device. The next future experiments will be agreed with centres where bigger cavies, like pigs are available, whose Basal Ganglia dimensions are more suitable for stereotactic neurosurgery in view of a future use in Humans. VI. CONCLUSION The proposed device is a first step towards a wearable biomedical monitoring system, mainly devoted to

[2] [3] [4] [5] [6] [7] [8] [9] [10]

[11] [12] [13] [14] [15] [16] [17]

E. Y. Uc, K. A. Follett, “Deep Brain Stimulation in Movement Disorders” Seminars in Neurology, Volume 27(2), pp. 170-182, April 2007. http://www.oxford-instruments.com/Pages/home.aspx Datasheet WPI-ISO80, World Precision Instruments. Available: http://www.wpiinc.com/pdf/ISO-80-IM-1100.pdf AA.VV., “Medical Applications Guide”, Texas Instruments. Available: http://focus.ti.com/lit/ml/slyb108d/slyb108d.pdf A. Moran, H. Bergman, Z. Israel and I. Bar-Gad. “Subthalamic nucleus functional organization revealed by parkinsonian neuronal oscillations and synchrony.” Brain 2008 131(12):3395-3409. C. Brigante, “Sviluppo di un’architettura Software/Hardware per l’acquisizione ed il trattamento di segnali biomedici”, Master degree disseration, DIEES, University Of Catania, Catania, Italy, 2008. http://www.tracopower.com/ P. L. Gildenberg, “Evolution of neuromodulation”, Stereotact Funz Neurosurg, 83 (2-3), 71-79, 2005. Datasheet ESDA6V14BC6: www.st.com/stonline/products/literature/ds/11647.pdf . Metting van Rijn, A.C., Peper, A. and Grimbergen, C.A., “Highquality recording of bioelectric events: Part 2 - Low-noise, lowpower multichannel ampliÞer design“, Medical and Biological Engineering and Computing, 29:433-440 (1991). B. Winter and J. Webster, “Driven-right-leg circuit design” IEEE Trans. Biomed. Eng., vol. BME-30, pp. 62–66, Jan. 1983. D. Prutchi, M. Norris, Design and Development of Medical Electronic Instrumentation: A Practical Perspective of the Design, Construction, and Test of Medical Devices, Wiley, December 2004. R. Mancini, Op Amps for Everyone, August 2002, Available: http://focus.ti.com/lit/an/slod006b/slod006b.pdf S. Franco, Design with Operational Amplifiers and Analog Integrated Circuits”, McGraw-Hill, 2002. CEI norm EN 60601-1, Part 1. Datasheet Keithley: http://www.keithley.com/data?asset=50827. http://www.ni.com/dataacquisition/usb/

131

Authorized licensed use limited to: UNIVERSITA DI CATANIA. Downloaded on August 18, 2009 at 05:50 from IEEE Xplore. Restrictions apply.