An Implantable Wireless Neural Interface System for ... - MDPI

sensors Article

An Implantable Wireless Neural Interface System for Simultaneous Recording and Stimulation of Peripheral Nerve with a Single Cuff Electrode Ahnsei Shon 1 , Jun-Uk Chu 1, *, Jiuk Jung 1 , Hyungmin Kim 2 1

2

*

ID

and Inchan Youn 2, *

Daegu Research Center for Medical Devices and Rehabilitation Engineering, Korea Institute of Machinery and Materials, 330, Techno Sunhwan-ro, Yuga-myeon, Dalseong-gun, Daegu 42994, Korea; [email protected] (A.S.); [email protected] (J.J.) Biomedical Research Institute, Korea Institute of Science and Technology, 5, Hwarang-ro 14-gil, Seongbuk-gu, Seoul 02792, Korea; [email protected] Correspondence: [email protected] (J.-U.C.); [email protected] (I.Y.); Tel.: +82-53-670-9105 (J.-U.C.); +82-2-958-5913 (I.Y.)

Received: 27 November 2017; Accepted: 15 December 2017; Published: 21 December 2017

Abstract: Recently, implantable devices have become widely used in neural prostheses because they eliminate endemic drawbacks of conventional percutaneous neural interface systems. However, there are still several issues to be considered: low-efficiency wireless power transmission; wireless data communication over restricted operating distance with high power consumption; and limited functionality, working either as a neural signal recorder or as a stimulator. To overcome these issues, we suggest a novel implantable wireless neural interface system for simultaneous neural signal recording and stimulation using a single cuff electrode. By using widely available commercial off-the-shelf (COTS) components, an easily reconfigurable implantable wireless neural interface system was implemented into one compact module. The implantable device includes a wireless power consortium (WPC)-compliant power transmission circuit, a medical implant communication service (MICS)-band-based radio link and a cuff-electrode path controller for simultaneous neural signal recording and stimulation. During in vivo experiments with rabbit models, the implantable device successfully recorded and stimulated the tibial and peroneal nerves while communicating with the external device. The proposed system can be modified for various implantable medical devices, especially such as closed-loop control based implantable neural prostheses requiring neural signal recording and stimulation at the same time. Keywords: electroneurogram; implantable medical device; MICS-band-based radio link; nerve cuff electrode; wireless power transmission

1. Introduction Neural prosthetic devices have been used for clinical purposes in various applications, including foot-drop correction [1], handgrip assistance [2], pain relief [3] and bladder control [4]. In the past, those kind of neural interface devices that use percutaneous connections between their electrodes and external devices not only restrict the subject’s mobility but also carry a serious risk of infection because of the lead wires penetrating the skin. In addition, the tethered cables are subject to considerable wear and often contaminate the neural signals due to external noise caused by the subject’s movements and nearby power lines [5–7]. To overcome these drawbacks, implantable wireless devices have attracted significant attention in recent years due to the elimination of infection and noise produced by the wires [8–12]. Although these devices have resolved the problems caused by connected wires, certain drawbacks still remain, including low-efficiency power transmission, short-range data communication

Sensors 2018, 18, 1; doi:10.3390/s18010001

www.mdpi.com/journal/sensors

Sensors 2018, 18, 1

2 of 26

with high power consumption and limited functionality (i.e., acting either as a neural signal recorder or as a stimulator). In this study, we address three major issues of implantable wireless neural interface system. The first issue is the low efficiency of the power transmission system and the limited power supply for an implanted device. Most implanted wireless devices are powered via either an inductive link or a non-rechargeable battery [6,9,10,12–19]. However, each system has its disadvantages: the former, a battery-free inductive link, requires very close distance between the implanted device and the external powering module, which restricts the movement of the subject wearing the device while providing low efficiency wireless power transmission; the latter, a non-rechargeable battery having its own life span, demands inevitable surgeries to replace an exhausted battery due to the limited life span. The second issue concerns the way data communication is performed over long distances with low power consumption. Inductive, optical and radio links have been used to establish a data link between the implanted device and the external device [8,9,16,17,20,21]. However, inductive and optical links require patients to remain an uncomfortable position for proper data communication because data transmission and receiving devices should be kept in very close distance. Also, inductive-link-based medical communication system can be easily affected by other communication systems in the same frequency band [22]. Meanwhile, radio links in the industrial, scientific and medical (ISM) radio bands such as IEEE 802.15.1-Bluetooth and IEEE 802.15.4-ZigBee produce large losses above 500 MHz because electromagnetic energy is absorbed by the high water content in body tissues [23]. The third issue is to devise an electronic circuit to share an electrode between the stimulator and the amplifier. If a concurrent neural stimulation and recording is required, the experimenter needs two distinct devices, one as a stimulator and one as an amplifier and two different types of electrode, one for recording and the other for stimulation; this typically causes difficulties when inserting all devices and electrodes into the limited inner space of a body. Additionally, large stimulus-related artifacts can easily contaminate small recorded neural signal of interest [24–29]. To overcome the aforementioned issues, this study proposes a new implantable wireless neural interface system that consists of a wireless power consortium (WPC)-compliant power transmission circuit, a medical implant communication service (MICS)-band-based radio link and cuff-electrode path controllers. First, the WPC-compliant power transmission circuit is able to wirelessly charge the rechargeable battery through transcutaneous wireless power transmission. This circuit can eliminate unnecessary battery replacements, guarantee a higher efficiency than conventional inductive power transmission systems and reduce battery-charging time. Next, the MICS-band-based radio link provides longer wireless data communication range with smaller power consumption compared to the previous studies using inductive or optical link. The Federal Communication Commission (FCC) regulated the MICS band at 402-405 MHz for implantable biotelemetry applications. By using the MICS band, operating distance between the implanted device and external device was increased, which is not possible with short-range magnetic coupling or optical link. The 402–405 MHz frequency band has superior signal propagation characteristics, that is, at the MICS band the RF signal in the human body could be less attenuated than other bands [30]. Therefore, the MICS-band-based RF transceivers consume a smaller current compared to other solutions that use the ISM radio bands. Additionally, unexpected interference with other services can be prohibited with particularly designed the channel bandwidth, output power level and duty cycle [22]. Finally, the newly devised switching circuit for the cuff-electrode path controller allows concurrent nerve stimulation and recording with a single cuff electrode while reducing the number of cables, cuff electrodes and electronic devices required. In this paper, we suggest a practical and affordable implantable wireless neural interface system for simultaneous neural signal recording and stimulation, which can be implemented using

Sensors 2018, 18, 1

3 of 26

Sensors 2018, 18, 1

3 of 26

readily available small form factor commercial off-the-shelf (COTS) components, instead of an application-specific integrated circuit (ASIC) that requires high cost andcomponents, design delays. The validity readily available small form factor commercial off-the-shelf (COTS) instead of an of the proposed design was confirmed by bench and in vivo using rabbit application-specific integrated circuit (ASIC) tests that requires high experiments cost and design delays. The models. validity of the proposed design was confirmed by bench tests and in vivo experiments using rabbit models.

2. Materials and Methods

2. Materials and Methods

Figure 1 shows a conceptual block diagram of the proposed system, which includes an implantable 1 shows a conceptual blockexternal diagramdevices of thewirelessly proposed transmit system, which includes device andFigure associated external devices. The commands andan power implantable device and associated external devices. The external devices wirelessly transmit to the implantable device. The implantable device stimulates peripheral nerves and records neural commands and power to the implantable device. The implantable device stimulates peripheral nerves signals through the cuff electrodes; it also wirelessly transmits the device’s status and recorded neural and records neural signals through the cuff electrodes; it also wirelessly transmits the device’s signals to the external devices. In addition, a personal computer (PC)-based graphical user interface status and recorded neural signals to the external devices. In addition, a personal computer (GUI)(PC)-based was also developed which specifies the commands required for stimulation and recording as graphical user interface (GUI) was also developed which specifies the commands well as displays stimulation waveforms andasrecorded The details of each and partrecorded and in vivo required for the stimulation and recording as well displays data. the stimulation waveforms experimental are part described theexperimental following sections. data. Theprocedures details of each and in in vivo procedures are described in the following sections. IMPLANTABLE DEVICE

SKIN

MICRO CONTROLLER

CUFF ELECTRODE

CUFF ELECTRODE PATH CONTROLLER

NEURAL AMPLIFIER NEURAL STIMULATOR

SPI CUFF ELECTRODE

CUFF ELECTRODE PATH CONTROLLER

2nd POWER REGULATOR

NEURAL AMPLIFIER NEURAL STIMULATOR

BATTERY

DATA

ADC WIRELESS DATA TRANSCEIVER

EXTERNAL DEVICES WIRELESS DATA TRANSCEIVER

MICRO CONTROLLER PC

DAC

1st POWER REGULATOR

USB CONTROLLER

WIRELESS POWER RECEIVER

POWER

WIRELESS POWER TRANSMITTER

USB POWER

Figure 1. Block diagram of the proposed implantable wireless device system.

Figure 1. Block diagram of the proposed implantable wireless device system.

2.1. Neural Signal Interface

2.1. Neural Signal Interface 2.1.1. Cuff Electrode

2.1.1. CuffThe Electrode cuff electrodes used in this research were developed by Center for BioMicroSystems at Korea Institute of Science and Technology (KIST) [31]. A revised quasi-tripolar configuration was The cuff electrodes used in this research were developed by Center for BioMicroSystems at Korea adopted to alleviate an imbalance from tissue impedance inside the cuff. When two middle Institute of Science and Technology (KIST) [31]. A revised quasi-tripolar configuration was adopted electrodes were placed at the center of the cuff electrode, a significant improvement in the to alleviate an imbalance from tissue impedance inside the cuff. When two middle electrodes were signal-to-interference ratio was achieved for the EMG signals and the stimulus artifacts when placed at the center of the cuff electrode, a significant improvement in the signal-to-interference ratio compared to a conventional tripolar cuff [5]. To fabricate a cuff electrode, a polyimide substrate and was achieved the EMG signals andused. the stimulus artifacts when compared to a conventional tripolar platinumfor electrode contacts were The polyimide substrate was self-biased to curl into a roll cuff [5]. To fabricate a cuff electrode, a polyimide substrate and platinum electrode contacts with an inner diameter. As shown in Figure 2, the fabricated cuff electrode has the following majorwere used.characteristics: The polyimide wasthe self-biased to curl1.5into roll withthe anmiddle inner diameter. 15 substrate mm between end electrodes, mma between electrodes As andshown a 1-mm2,diameter. The diameter of the cuff was designed to firmly wrap around tibial the in Figure the fabricated cuff electrode haselectrode the following major characteristics: 15 mmthe between and peroneal of the rabbit and to avoid damage excessive mechanical stresses on end electrodes, 1.5nerves mm between the middle electrodes andfrom a 1-mm diameter. The diameter of the the cuff nerve. Table 1 shows the physical characteristics of the fabricated cuff electrode. electrode was designed to firmly wrap around the tibial and peroneal nerves of the rabbit and to avoid damage from excessive mechanical stresses on the nerve. Table 1 shows the physical characteristics of the fabricated cuff electrode.

Sensors 2018, 18, 1

4 of 26

Sensors 2018, 18, 1

4 of 26

Figure 2. Nerve cuff electrode: (a) photographic image and (b) schematic image. Figure 2. Nerve cuff electrode: (a) photographic image and (b) schematic image. Table 1. Physical characteristics of the nerve cuff electrode. Table 1. Physical characteristics of the nerve cuff electrode.

Parameter Parameter Diameter of cuff Diameter of cuff Thickness of cuff Thickness of cuff Distance between end electrodes Distance between end electrodes Distance between middle electrodes Distance between middle electrodes Length of electrodeLength of electrode Width of electrode Width of electrode Impedance @ 1 kHzImpedance @ 1 kHz Charge delivery capacity Charge delivery capacity

Value Value 1 mm 1 mm 18 µm 18 µm 15 mm 15 mm 1.5 mm 1.5 mm 4 mm 4 mm 0.35 0.35 mm mm 1.2 kΩ 1.2 kΩ 2 22 µC/mm 2 22 µC/mm

2.1.2. 2.1.2. Neural Neural Amplifier Amplifier To Toacquire acquireneural neuralsignals, signals, the the cuff cuffelectrode electrode recording recording system system developed developed in in the the previous previous study study [5] [5] was modified to be suitable for implantable device. The neural amplifier was configured with a radio was modified to be suitable for implantable device. The neural amplifier was configured with a frequency interference (RFI) filter, AC filter, coupling a preamplifier, band pass filters, programmable radio frequency interference (RFI) ACfilters, coupling filters, a preamplifier, band pass filters, gain amplifiers (PGA) and a blanking circuit (see Figure 3). programmable gain amplifiers (PGA) and a blanking circuit (see Figure 3). A A precision, precision, low-noise low-noise instrumentation instrumentation amplifier amplifier (INA118, (INA118, Burr-Brown, Burr-Brown, Tucson, Tucson, AZ, AZ, USA) USA) was was used used as as the the preamplifier preamplifier and and its its gain gain was was fixed fixed at at 100. 100. The first band band pass pass filter filter and and the the first first PGA PGA followed followed the the preamplifier. preamplifier. Next, the second band pass filter and the second PGA were connected. The The band band pass pass filters filters consisted consisted of of aa second-order second-order Butterworth Butterworth high high pass pass filter filter with with aa 300-Hz 300-Hz cut-off cut-off frequency with a second-order Butterworth low pass a 5000-Hz frequency. frequencyinincascade cascade with a second-order Butterworth lowfilter passwith filter with a cut-off 5000-Hz cut-off Two dual matched PGAs (LTC6911, Linear Technology, Milpitas, CA, USA) were used to frequency. Two dual matched PGAs (LTC6911, Linear Technology, Milpitas, CA, USA) wereallow used selecting and gains each was with an ACwith coupling filter at the output of the band to allow various selectinggains various andlocated each was located an AC coupling filter atstage the output stage pass filters. of the band pass filters. The each channel waswas adjusted usingusing a three-bit parallelparallel interface to selecttovoltage The desired desiredgain gainof of each channel adjusted a three-bit interface select gains. Asgains. a result, overallthe amplifier wassystem designed have an adjustable gain betweengain 100 voltage As the a result, overallsystem amplifier wastodesigned to have an adjustable and 1,000,000 and a maximally flat magnitude response in the frequency range of 300 to 5000 between 100 and 1,000,000 and a maximally flat magnitude response in the frequency range of Hz, 300 where theHz, energy spectrum of the electroneurogram (ENG) signal is dominantly distributed. to 5000 where the energy spectrum of the electroneurogram (ENG) signal is dominantly Electrical stimulation generates stimulus artefacts, evoked ENG and electromyogram (EMG) distributed. signals, which continue for about 10 msstimulus after the artefacts, electrical stimulation [5,24–29]. Therefore, a blanking Electrical stimulation generates evoked ENG and electromyogram (EMG) circuit required to prevent signals: the stimulus artefacts and evoked ENGa signals,is which continue for acquiring about 10 unwanted ms after the electrical stimulation [5,24–29]. Therefore, and EMG signals [32]. In this study, single-pole, double-throw (SPDT) switches (ADG836L, Analog blanking circuit is required to prevent acquiring unwanted signals: the stimulus artefacts and Devices, Norwood, were usedIntothis construct blanking circuit. The operation time of the evoked ENG and MA, EMGUSA) signals [32]. study,the single-pole, double-throw (SPDT) switches blanking circuit could be set by a 16-bit timer a microprocessor (MSP430F1611, Texas Instruments, (ADG836L, Analog Devices, Norwood, MA, in USA) were used to construct the blanking circuit. The Dallas, TX, time USA) of from 10 ms with a resolution 1 ms. the timer selected of time, the operation the0 to blanking circuit could beof set byDuring a 16-bit in period a microprocessor devised blankingTexas circuitInstruments, disconnectedDallas, the amplified neural signal and blanking circuit’sofoutput (MSP430F1611, TX, USA) from 0 to 10tied ms the with a resolution 1 ms. with the the ground so that onlyof thetime, signal interest blanking was acquired. During selected period theofdevised circuit disconnected the amplified neural signal and tied the blanking circuit’s output with the ground so that only the signal of interest was acquired.

Sensors 2018, 18, 1

5 of 26

Sensors 2018, 18, 1 Sensors 2018, 18, 1

5 of 26 5 of 26

Figure 3. Block diagram of the proposed neural amplifier. Figure 3. Block diagram of the proposed neural amplifier. Figure 3. Block diagram of the proposed neural amplifier.

2.1.3. Neural Stimulator 2.1.3. Neural Neural Stimulator Stimulator 2.1.3. Figure 4 shows the voltage-controlled current source (VCCS) that was used for neural Figure 44shows thethe voltage-controlled current sourcesource (VCCS)(VCCS) that wasthat usedwas for neural stimulation. Figure shows voltage-controlled used neural stimulation. The current source should provide acurrent sufficiently large output impedance and afor minimized The current source should provide a sufficiently large output impedance and a minimized output stimulation. The error current source provide a sufficiently largeofoutput impedance and a minimized output current should beshould guaranteed over a wide range stimulation frequencies [33,34]. To currentcurrent error should be guaranteed over a wide range of range stimulation frequencies [33,34]. To maintain output error should be guaranteed over a wide of stimulation frequencies [33,34]. To maintain these conditions, a dual operational-amplifier current source was used as the current these conditions, a dual operational-amplifier current source was used as the current source. A 1µF maintain these conditions, a dual operational-amplifier current source was used as the current source. A 1µF coupling capacitor was used to prevent charge buildup at the electrode–tissue interface. coupling wascapacitor used to prevent charge buildup at thebuildup electrode–tissue interface. interface. source. A capacitor 1µF coupling was used to prevent charge at the electrode–tissue

Figure 4. Block diagram of the proposed voltage-controlled current source. Figure 4. Block diagram of the proposed voltage-controlled current source. Figure 4. Block diagram of the proposed voltage-controlled current source.

Also, symmetrical biphasic stimulation was used to prevent the tissue electrode damage [35–37]. Also, symmetrical biphasic stimulation was used the tissue electrode damage [35–37]. The symmetrical biphasic pulses were generated viatoaprevent digital-to-analogue converter (DAC) in the The symmetrical biphasic pulses were generated via a digital-to-analogue converter (DAC) the microprocessor and applied to the input of the VCCS. The generated biphasic pulses had a in 20-ms microprocessor and applied to the input of the VCCS. The generated biphasic pulses had a 20-ms

Sensors 2018, 18, 1

6 of 26

Also, symmetrical biphasic stimulation was used to prevent the tissue electrode damage [35–37]. Sensors 2018, 18, 1 6 of 26 The symmetrical biphasic pulses were generated via a digital-to-analogue converter (DAC) in the microprocessor and applied to the input of the VCCS. The generated biphasic pulses had a 20-ms period and and each each phase phase (i.e., (i.e., the the cathodal cathodal phase period phase and and the the anodal anodal phase) phase) had had aa fixed fixed duration duration of of 200 200 µs. µs. Therefore, the biphasic waveform with a fixed 50-Hz stimulation frequency was used throughout Therefore, the biphasic waveform with a fixed 50-Hz stimulation frequency was used throughout the the neural stimulation procedure. Contrary to stimulation a fixed stimulation and the duration, the neural stimulation procedure. Contrary to a fixed frequencyfrequency and duration, amplitude amplitude of the biphasic pulse could be set from 0 to 1500 µA in 20-µA increments. The of the biphasic pulse could be set from 0 to 1500 µA in 20-µA increments. The stimulation current stimulation current using could the be calculated using the following equation: could be calculated following equation: Vstim Istim Vstim . Istim = Rstim .

(1) (1)

Rstim During neural stimulation, ramp-up and ramp-down periods are essential to avoid sudden During neural stimulation, ramp-up periods arereactions essential [38]. to avoid sudden muscle contraction and relaxation, whichand mayramp-down cause spastic muscle To generate muscle relaxation, which may cause muscle reactions To generate diverse diversecontraction stimulationand waveforms with ramp-up and spastic ramp-down periods, the[38]. stimulation waveform stimulation waveforms with ramp-up and ramp-down periods, the stimulation waveform consisted of consisted of five different periods: a ready period, a ramp-up period, a hold period, a ramp-down five different a ready period, a ramp-up hold aperiod, a ramp-down period a rest period and aperiods: rest period. Each period could beperiod, set to ahave duration of 20 to 2000 msand in 20-ms period. Each(see period could increments Figure 5). be set to have a duration of 20 to 2000 ms in 20-ms increments (see Figure 5).

Figure 5. Stimulation waveform: (a) including ramp-up and ramp-down periods and (b) details of a Figure 5. Stimulation waveform: (a) including ramp-up and ramp-down periods and (b) details of a single pulse. single pulse.

2.1.4. Cuff-electrode Path Controller 2.1.4. Cuff-Electrode Path Controller A cuff-electrode path controller was designed to use one cuff electrode for recording and A cuff-electrode path controller was designed to use one cuff electrode for recording and stimulation. The cuff electrode’s path was determined by two analogue switches (TS3A24159, Texas stimulation. The cuff electrode’s path was determined by two analogue switches (TS3A24159, Texas Instruments, Dallas, TX, USA). Instruments, Dallas, TX, USA). As shown in Figure 6, when the control signal of the cuff-electrode path controller was set to As shown in Figure 6, when the control signal of the cuff-electrode path controller was set to “low”, the cuff electrode began to work with the amplifier. Conversely, when the control signal of “low”, the cuff electrode began to work with the amplifier. Conversely, when the control signal of the the cuff-electrode path controller was “high”, the cuff electrode began to work with the stimulator. cuff-electrode path controller was “high”, the cuff electrode began to work with the stimulator.

Sensors 2018, 18, 1

7 of 26

Sensors 2018, 18, 1

7 of 26

Figure 6. Block diagram of the proposed cuff-electrode path controller. Figure 6. Block diagram of the proposed cuff-electrode path controller.

2.2. Wireless Data Communication 2.2. Wireless Data Communication Two ZL70102 medical implant RF transceivers (Microsemi, Aliso Viejo, CA, USA) were used TwoMICS-band-compliant ZL70102 medical implant RF transceivers (Microsemi, Aliso CA, USA) were used the for for the wireless data communication. As anViejo, ultra-low-power device, the MICS-band-compliant wireless data communication. As an ultra-low-power device, the ZL70102 ZL70102 medical implant RF transceiver provides high-speed communication with raw data rates medical implant RFkbps. transceiver providesdevice high-speed communication with raw data rates of 800, 400, of 800, 400, or 200 The ZL70102 has three different power-consuming stages: a sleep or 200 kbps. The ZL70102 device has three different power-consuming stages: a sleep stage (100 nA), a stage (100 nA), a wake-up stage (380 nA) and a stage of high-speed transmitting or receiving (5 mA) wake-up stage (380 nA) and a stage of high-speed transmitting or receiving (5 mA) (ZL70102 Design (ZL70102 Design Manual) [39]. Two microprocessors (MSP430F1611, Texas Instruments, Dallas, TX, Manual) [39].used, Two microprocessors (MSP430F1611, Dallas, TX, USA)to were used,each one USA) were one in the implantable deviceTexas and Instruments, one in the external device, control in the implantable device and one in the external device, to control each wireless data transceiver using wireless data transceiver using the serial peripheral interface (SPI). For operation in the 400-MHz the serial peripheral operation in thehelical 400-MHz MICSfor band the 2.45-GHz ISMa MICS band and the interface 2.45-GHz(SPI). ISM For band, a dual-band antenna theand external device and band, a dual-band helical antenna for the external device and a wire whip antenna for the implantable wire whip antenna for the implantable device were used. device used. a session between the implantable device and the external device, a 2.45-GHz Towere establish To establish a session the implantable device and the externalmessages device, a 2.45-GHz wake-up transmitter in between the external device broadcasted wake-up to searchwake-up for the transmitter in the external device broadcasted wake-up messages to search for implantable device until it received a response from the implantable device. Whenthe the implantable implantable device until it received a response from the the external implantable device. When the device device received a wake-up message from device and responded to implantable the external device, received wake-up message from the external devicethe and responded device to the external wireless wireless aMICS-band link was established between implantable and the device, external device MICS-band link was established between the implantable device and the external device for wireless for wireless data transmission. data transmission. Figure 7 shows the block structure of the data packet for the wireless data transceiver used in Figure Both 7 shows the block of the dataimplantable packet for the wireless transceiver used in this this study. wireless data structure transceivers in the device and data the external device used the study. Both wireless data transceivers in the implantable device and the external device used the same same data packet, which was comprised of 31 blocks and the user data section consisted of 14 bytes. data packet, of 31the blocks and the usersent datacommands section consisted 14 bytes. Bydevice using By using thewhich givenwas usercomprised data sections, external device and theofimplantable the given user data sections, the external device sent commands and the implantable device transmitted transmitted the acquired neural signals. Figure 7c shows the user data section, organized for the acquired this neural signals. 7c parameters shows the user dataneural section, organized forthe commands; commands; data sectionFigure carried for the amplifier gain, blanking this time,data the section carried parameters neural amplifier stimulation amplitude andfor thethe stimulation period.gain, the blanking time, the stimulation amplitude and the Thestimulation amplified period. ENG signals were digitized with a sampling rate of 10 kHz per channel by an The amplified ENG signals were digitized sampling rate and of 10were kHz then per channel by for an internal analogue-to-digital converter (ADC) inwith the amicroprocessor packaged internal analogue-to-digital converter (ADC) in the microprocessor and were then packaged for transmission. Figure 7d shows the user data section designed for the acquired ENG signals. The user transmission. Figure 7dtwo shows the user section designed ENG The user data section included channels of data neural signals. Withinfor thethe 14acquired bytes, the oddsignals. bytes contained data section included channels of even neural signals. Within the 14 bytes, theof odd bytessignals. contained one channel of neural two signals and the bytes contained the other channel neural The one channel of neural signals and the even bytes contained the other channel of neural signals. 31 31 blocks of user data sections including recorded neural signals were accumulated into The a data blocks of user data sections including recorded neural signals were accumulated into a data packet packet and the data packet was sent from the implanted device to external device every 43.4 ms. and the data packet was sent from the implanted device to external device every 43.4 ms.

Sensors 2018, 18, 1

8 of 26

Sensors 2018, 18, 1

8 of 26 Preamble (48 Bits)

Header (130 Bits)

Data Block 1 (155 Bits)

Data Block 2

User Data (113 Bits = 14 Bytes + 1 Bit )

Amp Gain

Blanking Time

1

2

1 Byte

1 Byte

CH1 Amplitude

3

4

CRC (12 Bits)

CH1 Time

5

6

2 Byte

Data Block N+1

Data Block N

Data Block 30

Reed-Solomon (30 Bits)

Data Block 31

(b)

CH2 Amplitude

7

8

9

4 Byte

10

(a)

CH2 Time

11

2 Byte

12

13

14

(c)

13

14

(d)

4 Byte

CH1 ENG 7Bytes

1

2

3

4

5

6

7

8

9

10

11

12

CH2 ENG 7Bytes

Figure proposed data data packet packet and and its its structure: structure: (a) (a) data data packet, packet, (b) (b) data data block, block, Figure 7. 7. Block Block diagram diagram of of the the proposed (c) user data section for commands and (d) user data section for acquired neural signals. (c) user data section for commands and (d) user data section for acquired neural signals.

2.3. Wireless Power Transmission 2.3. Wireless Power Transmission Inefficient wireless power transmission devices waste energy and take a long time to charge Inefficient wireless power transmission devices waste energy and take a long time to charge their their rechargeable batteries. To minimize these weaknesses as well as to provide reliable and rechargeable batteries. To minimize these weaknesses as well as to provide reliable and transcutaneous transcutaneous wireless power transmission, Texas Instruments’ high efficiency wireless power wireless power transmission, Texas Instruments’ high efficiency wireless power transferring solution, transferring solution, known as Qi wireless charging, satisfying WPC standard was used. The WPC known as Qi wireless charging, satisfying WPC standard was used. The WPC developed an open developed an open interface standard that defines wireless power transfer using inductive charging. interface standard that defines wireless power transfer using inductive charging. Also, recent study Also, recent study regarding wireless power transmission for implantable devices revealed that the regarding wireless power transmission for implantable devices revealed that the induced magnetic field induced magnetic field using Qi wireless charging was well below the limits set by ISO 14117 using Qi wireless charging was well below the limits set by ISO 14117 evaluating the electromagnetic evaluating the electromagnetic compatibility (EMC) of active implantable cardiovascular devices [40]. compatibility (EMC) of active implantable cardiovascular devices [40]. An integrated wireless power supply receiver (BQ51013, Texas Instruments, Dallas, TX, USA) An integrated wireless power supply receiver (BQ51013, Texas Instruments, Dallas, TX, USA) was was used to receive power and a wireless power controller (BQ500211, Texas Instruments, Dallas, used to receive power and a wireless power controller (BQ500211, Texas Instruments, Dallas, TX, USA) TX, USA) was used to transmit power. The BQ51013 is a WPC-compliant, integrated, receiver IC for was used to transmit power. The BQ51013 is a WPC-compliant, integrated, receiver IC for wireless wireless power transfer in portable applications. It has a small package (1.9 mm × 3 mm) that is power transfer in portable applications. It has a small package (1.9 mm × 3 mm) that is suitable suitable for use in a compact implantable device. In addition, the BQ500211 controls to transmit for use in a compact implantable device. In addition, the BQ500211 controls to transmit external external power to the implantable device wirelessly. To save energy, the power transmitter stops power to the implantable device wirelessly. To save energy, the power transmitter stops supplying supplying power by entering a low-power standby mode when there is no need for power. The power by entering a low-power standby mode when there is no need for power. The transmitter and transmitter and receiver communicate with each other. For communication, various types of receiver communicate with each other. For communication, various types of communication packets communication packets are defined, such as identification and authentication packets, error packets, are defined, such as identification and authentication packets, error packets, control packets, end control packets, end power packets and power usage packets [41]. Unlike other inductive power power packets and power usage packets [41]. Unlike other inductive power transmission devices, transmission devices, WPC-compliant devices can avoid certain dangerous situations such as WPC-compliant devices can avoid certain dangerous situations such as unintended energy transfer unintended energy transfer to incompatible devices and reaching unacceptable temperature levels to incompatible devices and reaching unacceptable temperature levels in the power transferring in the power transferring device caused by metal objects by using identification communications device caused by metal objects by using identification communications and a metal object detection and a metal object detection function [42,43]. Therefore, users can reduce the time required for function [42,43]. Therefore, users can reduce the time required for charging a rechargeable battery in charging a rechargeable battery in the implantable device and improve safety by limiting wireless the implantable device and improve safety by limiting wireless power transmission when there are power transmission when there are some undesirable situations. some undesirable situations. Two different coils were fabricated for wireless power transmission. Unlike the wireless power Two different coils were fabricated for wireless power transmission. Unlike the wireless power transmitting coil detailed in the standards, certain customizations were required for the wireless transmitting coil detailed in the standards, certain customizations were required for the wireless power power receiving coil to be fitted into the small form factor of the implantable device. The receiving receiving coil to be fitted into the small form factor of the implantable device. The receiving coil was coil was designed to be as large as possible to maximize the overlapped area with the transmitting designed to be as large as possible to maximize the overlapped area with the transmitting coil while coil while also ensuring that it could be embedded in the implantable device. These customizations could have included a larger-diameter wire, the use of bifilar wire, or a printed circuit board (PCB) coil; for this design, however, bifilar wire was selected to fabricate the wireless power-receiving coil

Sensors 2018, 18, 1

9 of 26

also ensuring that it could be embedded in the implantable device. These customizations could have included a larger-diameter wire, the use of bifilar wire, or a printed circuit board (PCB) coil; for this design, however, bifilar wire was selected to fabricate the wireless power-receiving coil because it could provide a higher efficiency in the limited area than the larger-diameter wire or a PCB coil [42]. As a result, the number of turns in the wireless power receiving coil was set to 13 with an outer diameter of 23.6 mm. As a shielding material, a ferrite plate with a 2-mm thickness was attached to the bottom of the wireless power receiving coil to minimize the degradation of efficiency and prevent direct heat damage to susceptible components behind the wireless power receiving coil [42]. Other characteristics of the wireless power transmission coils are shown in Table 2. Table 2. Characteristics of the wireless power transmission coils. Parameter

Value

Wireless power transmitting coil Outer diameter Inner diameter Thickness Turns Layers Strands Inductance

43 mm 20 mm 2.1 mm 10 2 105 24.12 µH Wireless power receiving coil

Outer diameter Inner diameter Thickness Turns Layer Strands Inductance

23.6 mm 9.8 mm 0.3 mm 13 1 2 6.04 µH

2.4. In Vivo Experiments In vivo experiments were performed using adult male New Zealand white rabbits (2–3 kg) to record and stimulate the peripheral nerves. Throughout the experiments, the rabbits were kept and handled in accordance with the regulations of the Institutional Animal Care and Use Committee of Korea Institute of Science and Technology (KIST). An anesthetic mixed with Zoletil 50 (Virbac, Carros, France) and Rompun (Bayer, Seoul, Korea) was used for anesthesia. After removing the hair on the thigh of the animal, approximately an 3 cm incision on the rear side of the left hind limb was made to expose the sciatic nerve through the avascular fascicular junction [44]. Through the incision, the tibial and peroneal nerves were gently exposed; then, the two nerves were carefully separated to ensure space for the installation of the two cuff electrodes (see Figure 8). Finally, the cuff electrodes, pre-soaked in saline, were wrapped around the tibial and peroneal nerves, respectively (see Figure 9a). Another incision was made on the back skin of the rabbit to insert the implantable device (see Figure 9b). After inserting the implantable device under the back skin, all incision sites were closed. To remove all voluntary and reflex responses, the rabbit was under anesthesia during experiments. Supplemental doses of the anesthetic were administered every 20 min throughout the surgery and the experiments. During the experiments, the rabbit was placed on an apparatus and its ankle was fastened on a rotatable manipulator. Two apparatuses were devised for the in vivo experiments: one was equipped with a high-torque gear servomotor (HS-7954SH, Hitec, Shoeburyness, UK) to consistently provide passive movements of the rabbit’s ankle in the sagittal plane; the other was designed to stably support the evoked movements of the rabbit’s ankle via electrical stimulation that was generated from the implantable device. The angle of the rabbit’s ankle joint was measured using a goniometer (SG65, Biometrics), a

Sensors 2018, 18, 1

10 of 26

power and amplifier module (Electro Goniometer Interface Box, Motion Lab Systems, Baton Rouge, LA, USA) and a data acquisition card (NI PCI-6034E, National Instruments, Austin, TX, USA). Sensors 2018, 18, 1 10 of 26 Sensors 2018, 18, 1

10 of 26

Figure 8. Implantation sites of the cuff electrodes and definition of the ankle joint angles. Figure Figure 8. 8. Implantation Implantation sites sites of of the the cuff cuff electrodes electrodes and and definition definition of of the the ankle ankle joint joint angles. angles.

Figure 9. Surgical procedure for implant placement: (a) cuff electrodes wrapped around the tibial Figure 9. Surgical procedure for implant placement: (a) cuff wrapped around the tibial and peroneal nerves, (b) implantable device inserted theelectrodes back skin of a rabbit. Figure 9. Surgical procedure for implant placement: (a)under cuff electrodes wrapped around the tibial and and peroneal nerves, (b) implantable device inserted under the back skin of a rabbit. peroneal nerves, (b) implantable device inserted under the back skin of a rabbit.

2.4.1. Neural Signal Recording 2.4.1. Neural Signal Recording 2.4.1. To Neural Signal evaluate theRecording recording performance of the proposed system in the in vivo experiments, the To evaluate the recording performance of the proposed system in theidentical in vivo experiments, the priorTo studies [45,46] were performance modified as follows. A system series in ofthe ramp-and-hold, evaluate the recording of the proposed in vivo experiments, the prior studies [45,46] were were modified Arotate series ofrabbit’s identical ramp-and-hold, flexion-extension used astoAfollows. passively the ankle joint using a prior studies [45,46]movements were modified as follows. series of identical ramp-and-hold, flexion-extension flexion-extension movements were used to passively rotate the rabbit’s ankle joint using a high-torque gear servomotor. Fullflexion, neutral and full-extension positions were set at 70°, 100° movements were used to passively rotate the rabbit’s ankle joint using a high-torque gear servomotor. high-torque gear servomotor. Fullflexion, neutral and full-extension positions were set at 70°, 100° ◦ and and 130° (seeneutral Figureand 8) and each position had a were holding 2, 0.5 and130 2 ◦s,(see respectively. An ankle Fullflexion, full-extension positions set time at 70◦of , 100 Figure 8) and each and 130° (see Figure 8) and each position had a holding time of 2, 0.5 and 2 s, respectively. An ankle ◦ excursion of 60° was performed from full-extension to full-flexion and vice versa. A constant angular position had a holding time of 2, 0.5 and 2 s, respectively. An ankle excursion of 60 was performed excursion 60° was performed from full-extension to full-flexion vice versa.acquired A constant angular velocity ofof60°/s was during movements. The ENGand signals with a gain from full-extension to maintained full-flexion and vicethe versa. A constant angular velocitywere of 60◦ /s was maintained velocity of 60°/s was maintained during the movements. The ENG signals were acquired with a gain of 100 dB the blanking set to 0 ms. All parameters for of generating electrical stimulation during theand movements. The time ENGwas signals were acquired with a gain 100 dB and the blanking time of 100 dB and the blanking time was set to 0 ms. All parameters for generating electrical stimulation were 0 toAll only record muscle afferent electrical activities stimulation caused by passive without the was setset to to 0 ms. parameters for generating were setankle to 0 tomotions only record muscle were set to 0 to only record muscle afferent activities caused by passive ankle motions without the electrical stimulation. afferent activities caused by passive ankle motions without the electrical stimulation. electrical stimulation. 2.4.2. Simultaneous SimultaneousNeural NeuralSignal SignalRecording Recordingand andStimulation Stimulation 2.4.2. 2.4.2. Simultaneous Neural Signal Recording and Stimulation Another in invivo vivoexperiment experiment was was designed designed toto evaluate evaluate the the comprehensive comprehensive performance performance of of Another Another in vivo experiment was designed to evaluatebipolar the comprehensive performance of simultaneous recording and stimulation. stimulation. In this this experiment, experiment, pulses with with aa 400-µs 400-µs duration, simultaneous recording and In bipolar pulses duration, simultaneous recording and stimulation. thisfor experiment, bipolar pulsesand with a 400-µs duration,a 20-msperiod period and and 300-µA amplitude wereIn used tibial 20-ms 300-µA tibial nerve nerve stimulation stimulation andbipolar bipolarpulses pulseswith with 20-ms period and 300-µA amplitude were used for tibial nerve stimulation andnerve bipolar pulses with a period and 500-µA amplitude were used for for peroneal stimulation. The a400-µs 400-µsduration, duration,20-ms 20-ms period and 500-µA amplitude were used peroneal nerve stimulation. 400-µs duration, 20-ms period and 500-µA amplitude were used for peroneal nerve stimulation. The ENGENG signals werewere acquired with awith gain aofgain 100 dB time periods applied to applied the both The signals acquired of and 100 the dB same and the same timewere periods were ENG signals were acquired with a gain of 100 dB and the same time periods were applied to the stimulation waveforms waveforms for the tibialfor andthe peroneal nerves: readynerves: (1000 ms), ramp-up (200 ramp-up ms), both hold to the both stimulation tibial and peroneal ready (1000 ms), stimulation waveforms (200 for the peroneal ready (1000 ms), ramp-up (200 ms),[5,24– hold (1000 ms), ramp-down ms)tibial and and rest ms)nerves: periods. based on previous (200 ms), hold (1000 ms), ramp-down (200(1500 ms) and rest (1500Then, ms) periods. Then, basedstudies on previous (1000 (200was ms) set andtorest (1500 periods. Then,artefacts, based onevoked previous studies 29,47],ms), theramp-down blanking time 8 ms to ms) prevent stimulus ENG and[5,24– EMG 29,47], the blanking time was set to 8 ms to prevent stimulus artefacts, evoked ENG and EMG signals caused by electrical stimulation. The ENG signals were recorded starting from the neutral signals by electrical The ENG signals were recorded starting from the neutral positioncaused of the ankle angle atstimulation. 100°. position of the ankle angle at 100°.

Sensors 2018, 18, 1

11 of 26

studies [5,24–29,47], the blanking time was set to 8 ms to prevent stimulus artefacts, evoked ENG and EMG signals caused by electrical stimulation. The ENG signals were recorded starting from the neutral Sensors 2018, 11 of 26 position of 18, the1 ankle angle at 100◦ . Sensors 2018, 18, 1 11 of 26

3. Results Results 3. 3. Results

10 108 86 64 42 20 -02 --24 --46 - -68 --10 8 -10 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

10 10 8 8 6 6 4 4 2 2 0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 Frequency 3 4 5 6 [kHz] 7 8 9 10

Amplitude [µV] Amplitude [µV]

Amplitude [µV] Amplitude [µV]

3.1. 3.1. Neural Neural Signal Signal Interface Interface 3.1. Neural Signal Interface 3.1.1. 3.1.1. Neural Neural Amplifier Amplifier 3.1.1. Neural Amplifier An An evaluation evaluation circuit circuit studied studied in in [5] [5] was was used used to to simulate simulate the the differential differential and and common-mode common-mode An evaluation circuit studied in [5]To was used toperformance simulate theofdifferential and common-mode voltages as inputs to the neural amplifier. verify the the devised neural amplifier, the voltages as inputs to the neural amplifier. To verify the performance of the devised neural amplifier, voltages as inputs to the neural amplifier. To verify the performance of the devised neural amplifier, amplified output and frequency response of the overall system were measured. A sinusoidal the amplified output and frequency response of theamplifier overall amplifier system were measured. A the amplified outputwith anda frequency response and of the overall amplifier system were measured. A signal was provided 1-V peak amplitude 1.5-kHz frequency and was attenuated the level sinusoidal signal was provided with a 1-V peak amplitude and 1.5-kHz frequencytoand was sinusoidal signalthis was provideddifferential with a 1-V peakrepresented amplitude the and 1.5-kHz frequency and was of a 10-µV peak; attenuated voltage ENG signal. The common-mode attenuated to the level of a 10-µV peak; this attenuated differential voltage represented the ENG attenuated to the level of a 10-µV peak; this attenuated differential voltage represented thewith ENGa voltage was assumed to be the power-line interference. A sinusoidal signal was provided signal. The common-mode voltage was assumed to be the power-line interference. A sinusoidal signal. The common-mode voltage was assumed to be the power-line interference. A sinusoidal 1-V peak amplitude frequency without Figure without 10 showsattenuation. the output Figure signal signal was providedand witha a60-Hz 1-V peak amplitude andattenuation. a 60-Hz frequency signal was provided with a 1-V peak amplitude and a 60-Hz frequency without attenuation. Figure of neural in theoftime domainamplifier and the in frequency signals generated 10 the shows the amplifier output signal the neural the timedomain domainwith and test the frequency domain 10 shows the output signal of 10-µV, the neural amplifier in thevoltage time domain and the and frequency domain by the evaluation circuit. The 1.5-kHz differential was amplified the 1-V, 60-Hz, with test signals generated by the evaluation circuit. The 10-µV, 1.5-kHz differential voltage was with test signalsvoltage generated by the evaluation circuit. The 10-µV, 1.5-kHz differential voltage was common-mode was filtered out. amplified and the 1-V, 60-Hz, common-mode voltage was filtered out. amplified and the 1-V, 60-Hz, common-mode voltage was filtered out.

Time [ms] Time(a) [ms] (a)

Frequency (b) [kHz] (b)

Figure 10. 10. Measured Measured output output signal: signal: (a) (a) in in the the time time domain domain and and (b) (b) in in the the frequency frequency domain. domain. Figure Figure 10. Measured output signal: (a) in the time domain and (b) in the frequency domain.

As shown in Figure 11, the frequency response was measured with a dynamic signal analyser As shown shown in in Figure Figure 11, 11, the frequency frequency response response was was measured measured with with a dynamic dynamic signal signal analyser analyser As (35670A, Agilent, Santa Clara, CA, USA). The −3 dB bandwidth was found to be between 237–5478 (35670A, Agilent, 3 dB was found to to be be between 237–5478 Hz. (35670A, Agilent, Santa SantaClara, Clara,CA, CA,USA). USA).The The−−3 dBbandwidth bandwidth was found between 237–5478 Hz. These results were obtained when the neural amplifier’s total gain was 100 dB. These results werewere obtained when the neural amplifier’s totaltotal gaingain was was 100 dB. Hz. These results obtained when the neural amplifier’s 100 dB.

Gain[dB] Gain[dB]

120 120 100 100 80 80 60 60 40 40 20 20 1 1

10 10

100

1k

100 1k Frequency[Hz] Frequency[Hz]

10k 10k

Figure 11. Frequency response of the overall amplifier system. Figure Figure 11. 11. Frequency Frequency response response of of the the overall overall amplifier amplifier system. system.

The blanking circuit was used to selectively record signals so that it could prevent acquiring The blanking circuit was used to selectively record signals so that it could prevent acquiring The blanking was used to record so that itwith could acquiring unwanted signals. circuit The performance of selectively the blanking circuitsignals was evaluated theprevent same signal used unwanted signals. The performance of the blanking circuit was evaluated with the same signal used unwanted signals. The of the blanking circuit was evaluated with the signal used for the verification of performance the neural amplifier and the blanking control signal wassame generated by a for the verification of the neural amplifier and the blanking control signal was generated by a function generator with a duration of 500 µs. As shown in Figure 12, the output signal was function generator with a duration of 500 µs. As shown in Figure 12, the output signal was grounded when the blanking control signal was in its ON state. Table 3 shows the summarized grounded when the blanking control signal was in its ON state. Table 3 shows the summarized characteristics of the neural amplifier. characteristics of the neural amplifier.

Sensors 2018, 18, 1

12 of 26

for the verification of the neural amplifier and the blanking control signal was generated by a function generator with a duration of 500 µs. As shown in Figure 12, the output signal was grounded when the blanking control signal was in its ON state. Table 3 shows the summarized characteristics of the neural amplifier.

Sensors 2018, 18, 1

12 of 26

Sensors 2018, 18, 1

12 of 26

Figure 12. Measured output signal blanked during a 500-ms control signal. Figure 12. Measured output signal blanked during a 500-ms control signal. Table 3. Electrical characteristics of the neural amplifier. Table 3. Electrical characteristics of the neural amplifier. Figure 12. Measured output signal blanked during a 500-ms control signal. Parameter Value

Parameter Number of channels 2 amplifier. Value Table 3. Electrical characteristics of the neural Number channels 2 −3 dBof bandwidth 237–5478 Hz Parameter ValueHz −3Gain dB bandwidth 237–5478 38–123 dB Number of channels 2 38–123 dB Gain Noise @300 Hz~5 kHz, Gain 100 dB 792.47 nV −3 dB bandwidth 237–5478 Noise @300 Hz~5 kHz, Gain 100 dB 792.47Hz nV Sampling rate 10 kHz 38–123 dB SamplingGain rate 10 kHz Noise @300 Hz~5 kHz, Gain 100 dB 792.47 nV Blanking timetime 0–10 ms Blanking 0–10 ms Sampling rate

10 kHz 0–10 ms

Blanking time 3.1.2. Neural Neural Stimulator

Neuralthe Stimulator To 3.1.2. evaluate performance of the devised VCCS, a 1-kΩ 1-kΩ load resistor was connected to the output of of the the VCCS. The output current of the VCCS converted to was a voltage across the load The current VCCS was was converted a voltage across the load resistor ToVCCS. evaluate theoutput performance of of thethe devised VCCS, a 1-kΩ loadtoresistor connected to the resistor and resistor voltage was measured andwas compared totothe VCCS input the voltage. of load the VCCS. Thewas output current of thecompared VCCS converted ainput voltage across load and theoutput loadthe resistor voltage measured and to the VCCS voltage. resistor the load voltage was measured and compared to thethrough VCCS input Figure 13 and shows theresistor converted current stimulation waveforms thevoltage. load resistor with waveforms Figure 13 shows the converted current stimulation waveforms through the load resistor with various input voltages generated generated from the DAC of the microprocessor. It was possible to generate various input voltages generated from the DAC of the microprocessor. It was possible to generate precise biphasic, constant-current pulses for stimulation with minimal errors. Table 44 shows the constant-current precise biphasic, constant-current pulses for stimulation with minimal errors. Table Table 4 shows the summarized characteristics of the neural stimulator. neural stimulator. summarized characteristics of the neural stimulator. 800 800 240µA 240µA 360µA 360µA 480µA 480µA 600µA 600µA 720µA 840µA 720µA

600 600 400

400

840µA

200

960µA 960µA 1080µA 1080µA 1200µA 1200µA 1320µA 1320µA 1440µA

1440µA

200 0

0

- 200

- 200- 400 - 400- 600 - 600

- 800

0

100

200

300

400

500 600

700

800

900 1000

Time [µs] - 800

0 100VCCS 200 output 300 current 400 500 600 700 900 1000 Figure 13. Converted waveforms with800 a 1 kΩ load resistor.

Figure 13. Converted VCCS output current waveforms with a 1 kΩ load resistor. Time [µs]

Figure 13. Converted VCCS output current waveforms with a 1 kΩ load resistor.

Sensors 2018, 18, 1

13 of 26

Sensors 2018, 18, 1

13 of 26

Table 4. Electrical characteristics characteristics of of the the neural neural stimulator. Table 4. Electrical stimulator.

Parameter Parameter Number of channels Number of channels Amplitudeofofbiphasic biphasic pulse Amplitude pulse Duration and anodal phases Durationofofcathodal cathodal and anodal phases Period of biphasic pulse Period of biphasic pulse Stimulation waveform Stimulation waveform

Value Value 2 2 0–15000–1500 µA µA 200 µs 200 µs 20 ms 20 ms ready, ramp-up, hold, ramp-down, rest ready, ramp-up, hold, ramp-down, rest

3.1.3. Cuff-electrode Cuff-electrode Path Path Controller Controller 3.1.3. Two SPDT Texas Instruments, Dallas, TX, TX, USA) USA) were were used used to to Two SPDT analogue analogue switches switches (TS3A24159, (TS3A24159, Texas Instruments, Dallas, implement the cuff-electrode path controller. To evaluate the function of the cuff-electrode path implement the cuff-electrode path controller. To evaluate the function of the cuff-electrode path controller, conductivity and output For the the conductivity test, the the resistance controller, conductivity and output signals signals were were measured. measured. For conductivity test, resistance between the inputs and outputs was measured while changing the control signal of the cuff-electrode between the inputs and outputs was measured while changing the control signal of the path controller. Approximately 0 Ω was measured for each connection when the inputs and outputs cuff-electrode path controller. Approximately 0 Ω was measured for each connection when the were by the control signals. inputsconnected and outputs were connected by the control signals. Next, the output characteristics thethe cuff-electrode pathpath controller were were verified with the circuit Next, the output characteristicsof of cuff-electrode controller verified with the connections shown in Figure 14 while changing the control signal. When the cuff-electrode path circuit connections shown in Figure 14 while changing the control signal. When the cuff-electrode controller was working with the neural with a control of “low”, peak, path controller was working with the amplifier neural amplifier with asignal control signal aof1.5-kHz, “low”, 1.5-V a 1.5-kHz, sinusoidal signal was connected between pins 7 and 6 and the output signal was measured between 1.5-V peak, sinusoidal signal was connected between pins 7 and 6 and the output signal was pins 4 and 5between (see Figure Conversely, when the cuff-electrode path controller was working with measured pins14a). 4 and 5 (see Figure 14a). Conversely, when the cuff-electrode path the neural stimulator with a control signal of “high”, the biphasic pulses were connected controller was working with the neural stimulator with a control signal of “high”, the between biphasic pins 1 and and the output signal was measured pins 7 and 6 (see Figure between 14b). From the pulses were2 connected between pins 1 and 2 and between the output signal was measured pins 7 experimental results, it was confirmed that the cuff-electrode path was successfully shared between and 6 (see Figure 14b). From the experimental results, it was confirmed that the cuff-electrode path the neural amplifier and between stimulator onamplifier the control signal. was successfully shared thebased neural and stimulator based on the control signal.

Figure 14. Cont.

Sensors 2018, 18, 1 Sensors 2018, 18, 1

14 of 26 14 of 26

Figure 14. Circuit connections used to evaluate the cuff-electrode path controller: (a) for neural Figure 14. Circuit connections used to evaluate the cuff-electrode path controller: (a) for neural amplifier and (b) for neural stimulator. amplifier and (b) for neural stimulator.

3.1.4. Stimulation Waveforms and Control Signals 3.1.4. Stimulation Waveforms and Control Signals Figure 15 shows an example of the generated stimulation waveforms for the VCCS inputs and Figure 15 shows an example of the generated stimulation waveforms for the VCCS inputs and the the control signals for the electrode path controllers and blanking circuits. These signals were control signals for the electrode path controllers and blanking circuits. These signals were measured measured from the output ports of the microprocessor and their parameters were wirelessly from the output ports of the microprocessor and their parameters were wirelessly specified by the specified by the designed GUI software and external device. designed GUI software and external device. The upper portion of Figure 15 shows the signals measured over 10 s. The lower portions of The upper portion of Figure 15 shows the signals measured over 10 s. The lower portions of Figure 15 show enlargements of the upper portion of Figure 15, where all signals were captured Figure 15 show enlargements of the upper portion of Figure 15, where all signals were captured synchronously with a single biphasic stimulation pulse. In this example, to stimulate the rabbit’s synchronously with a single biphasic stimulation pulse. In this example, to stimulate the rabbit’s nerve, nerve, biphasic stimulation waveforms including ramp-up and ramp-down periods were produced biphasic stimulation waveforms including ramp-up and ramp-down periods were produced and each and each pulse in the biphasic stimulation waveform had a 400-µs duration. The same duration was pulse in the biphasic stimulation waveform had a 400-µs duration. The same duration was applied applied to the electrode path control signals to allow the cuff electrodes to only work with the to the electrode path control signals to allow the cuff electrodes to only work with the stimulator stimulator during this time period and to work with the amplifier during the other time period. As during this time period and to work with the amplifier during the other time period. As shown in the shown in the lower portion of Figure 15, the electrode path control signal had the same 20-ms lower portion of Figure 15, the electrode path control signal had the same 20-ms period as the biphasic period as the biphasic stimulation pulse and maintained as “high” for the fixed duration of 400 µs stimulation pulse and maintained as “high” for the fixed duration of 400 µs when only the biphasic when only the biphasic stimulation pulse was generated. Each blanking control signals began from stimulation pulse was generated. Each blanking control signals began from the start of the biphasic the start of the biphasic stimulation pulse and was maintained for 4 ms. Although the blanking stimulation pulse and was maintained for 4 ms. Although the blanking control signals can prevent control signals can prevent stimulus artefacts, evoked ENG and EMG signals from affecting one stimulus artefacts, evoked ENG and EMG signals from affecting one channel, these unwanted signals channel, these unwanted signals still remain on the opposite channel. Therefore, an additional still remain on the opposite channel. Therefore, an additional blanking control signal was also applied blanking control signal was also applied to the opposite channel at the same time with the same to the opposite channel at the same time with the same duration. duration.

Sensors 2018, 18, 1 Sensors 2018, 18, 1

15 of 26 15 of 26

Figure 15. Stimulation waveforms and control signals for the cuff-electrode path controllers and

Figure 15. Stimulation waveforms and control signals for the cuff-electrode path controllers and blanking circuits. blanking circuits.

3.2. Wireless Data Communication

3.2. Wireless Data Communication To evaluate the reliable maximum distance of the wireless data communication, sinusoidal signals were the applied to maximum the implantable device andwireless were monitored through the sinusoidal devised GUI To evaluate reliable distance of the data communication, signals program while increasing the distance between the implantable device and the external device. The were applied to the implantable device and were monitored through the devised GUI program while implantable device was placed under subcutaneous tissue after hermetic packaging. During this increasing the distance between the implantable device and the external device. The implantable experiment, it was determined whether the received data were identical to those applied to the device was placed under subcutaneous tissue after hermetic packaging. During this experiment, implantable device. The established session was maintained at up to 1.98 m and the sinusoidal

it was determined whether the received data were identical to those applied to the implantable device. The established session was maintained at up to 1.98 m and the sinusoidal signals were acquired without distortion. The current consumption of the MICS-band-based RF transceiver was

Sensors 2018, 18, 1

16 of 26

measured by using a current probe (TCP0030A, Tektronix, Beaverton, OR, USA) in different modes: data receiving mode (4.7 mA) and data transmitting mode (5.6 mA). The power consumption of the RF transceiver in Table 5 was calculated when 3.3 V was supplied. Other characteristics for the wireless data communication are shown in Table 5. Table 5. Characteristics of the wireless data communication. Parameter

Value

Radio frequency Data rate Modulation Power Consumption Operating distance

402–405 MHz (MICS band) 400 kbps FSK, bidirectional 15.51 mW (Receiving mode) 18.48 mW (Transmission mode) ~1.98 m

3.3. Wireless Power Transmission To verify the performance of the devised wireless power transmission function, its efficiency and maximum operating distance were investigated. The efficiency of the wireless power transmission system was calculated using the ratio of the output power to the input power. Next, the maximum distance was measured by increasing the distance of the implantable device and the external wireless power transmission module until there was no wireless power transmission. The implantable device was placed under subcutaneous tissue. From the outside, the external wireless power transmitting module was positioned close to the implantable device and the output power (transmitting power) and the input power (receiving power) were measured to calculate the efficiency of the devised wireless power transmission. The efficiency of the configured wireless power transmission device was 67% and the wireless power receiving and transmitting procedures were found to operate up to a maximum distance of 11 mm. Table 6 lists the characteristics of the wireless power transmission. Table 6. Characteristics of the wireless power transmission. Parameter Wireless power solution Switching frequency Input power of wireless power transmitter Output power of wireless power receiver Efficiency Operating distance

Value WPC V1.1 110–205 kHz 3.73 W (5 V, 0.746 A) 2.5 W (5 V, 0.5 A) 67% ~11 mm

3.4. Assembled Implantable Device Figure 16a shows the disassembled implantable device. The implantable device was implemented in two distinct parts: an analogue circuit for the neural amplifier and stimulator and a digital circuit for wireless data communication and wireless power receiving. The analogue and digital circuits were intentionally separated to protect the analogue paths from digital interference and power-supply noise. As shown in Figure 16b, the complete implantable device was assembled in a layered manner by stacking the analogue circuit, rechargeable battery, digital circuit with its antenna, shielding material and power receiving coil. For packaging the implantable device and two cuff electrodes, the bio-compatible PolyJet photopolymer (MED610) that is a rigid medical rapid prototyping material was used. After packaging the assembled implantable device with MED610, we coated the device with dimethylsiloxane (PDMS, sylgard 184, DowCorning, Midland, MI, USA) for in vivo experiments. The base and the curing agent were mixed in 10 to 1 mass ratio. A vacuum chamber was used to eliminate bubbles for 20 min. The curing process was conducted for 1.5 h in a pre-heated oven at 60 ◦ C. The cuff electrodes were encapsulated by closures. In the proposed system, a stable electrical connection and strain relief were accomplished by means of a closure that enveloped the cuff electrodes, lead wires

Sensors 2018, 18, 1

17 of 26

Sensors 2018, 18, 1

17 of 26

and nerve bundle. The closures were comprised of a base plate and slider plate of the rigid medical for 1.5 h in a pre-heated oven at 60 °C. The cuff electrodes were encapsulated by closures. In the rapid prototyping material. To verify the hermetically sealed implantable device, the implantable proposed system, a stable electrical connection and strain relief were accomplished by means of a device was immersed in 0.9%the saline a week.lead During wireless and closure that enveloped cuff for electrodes, wiresthis andtime, nerve bundle.data The communication closures were wirelesscomprised power transmission established once day. The disabled devices in the were not of a base plate were and slider plate of the rigida medical rapid prototyping material. Totest verify the for hermetically sealed implantable device, implantable device was immersed in 0.9% salinewhich for implanted animal experiments. Figure 16cthe shows the encapsulated implantable device, was a week. During this time, wireless data communication and wireless power transmission were 28 mm × 33 mm × 12 mm and 21 g. The implantable device consumed 6 mA in the ready stage, 25 mA established once a day. The disabled devices in the test were not implanted for animal experiments. during the only-recording stage and 33 mA during the simultaneous-recording-and-stimulation stage. Figure 16c shows the encapsulated implantable device, which was 28 mm × 33 mm × 12 mm and 21 g. The operating times ofdevice the implantable device to be 185the h (ready), 44.4 hstage (recording The implantable consumed 6 mA in thewere readyexpected stage, 25 mA during only-recording only) and h (recording stimulation) when the fully charged battery (3.7 V, 1.11 was used and33.6 33 mA during theand simultaneous-recording-and-stimulation stage. The operating timesWh) of the were(see expected be 185 h (ready), 44.4 h (recording only) and 33.6 h (recording withoutimplantable additionaldevice charging Tableto7). and stimulation) when the fully charged battery (3.7 V, 1.11 Wh) was used without additional charging (see Table 7). 7. Characteristics of the assembled implantable device. Table Table 7. Characteristics of the assembled implantable device.

Parameter

Parameter Power consumption

Power consumption

Value

Value 19.9 mW (ready) 82.5 mW (recording only) 19.9 mW (ready) 108.9 mW only) (stimulation and recording) 82.5 mW (recording

108.9 mW (stimulation 185 h (ready)and recording) Battery Operational hours after fully charged185 h (ready) 44.4 h (recording only) Battery Operational hours after 33.6 h (stimulation and recording) 44.4 h (recording only) fully charged

Battery charging time

Battery charging time Total dimension Total dimension Total weight Total weight

33.6 h (stimulation 1–1.2 h and recording) 1–1.2 h 28 mm × 33 mm × 12 mm 28 mm × 33 mm × 12 mm 21 g 21 g

Figure 16. Photograph of the implantable device: (a) disassembled view and key components of the

Figure 16. Photograph of the implantable device: (a) disassembled view and key components of the implantable device, (b) assembled view and (c) encapsulated implantable device and cuff electrodes implantable device, (b) assembled view and(MED610). (c) encapsulated implantable device and cuff electrodes by by bio-compatible PolyJet photopolymer bio-compatible PolyJet photopolymer (MED610).

Sensors 2018, 18, 1 Sensors 2018, 18, 1

18 of 26 18 of 26

3.5. In In Vivo Experiments Performance of Neural Signal Recording 3.5.1. Performance

Ankle Angle [degree]

Figure 17 shows the recorded joint angle and ENG signals during the passive movements of The muscle muscle afferent afferent activity activity from the tibial nerve showed the highest amplitude the rabbit’s ankle. The (70◦ ). Conversely, Conversely, the the muscle muscle afferent afferent activity activity from from the the peroneal nerve showed the at full-flexion (70°). ◦ ). The ENG signals were properly amplified and filtered with highest amplitude amplitudeatatfull-extension full-extension(130 (130°). The ENG signals were properly amplified and filtered a higha signal-to-noise ratio ofratio 5.46of dB5.46 and dB were then wirelessly transmitted at 10 kHzatsampling with high signal-to-noise and were then wirelessly transmitted 10 kHz rate from rate the from implantable device todevice the external device. device. The recorded ENG signals from from both sampling the implantable to the external The recorded ENG signals peroneal and and tibialtibial nerves corresponded to to thethe results [44–46]. both peroneal nerves corresponded resultsobtained obtainedininthe the previous previous studies [44–46]. neural signal recording with the passive movement of the rabbit’s ankle, the stimulation During neural thethe cuff-electrode path controller. The The obtained neural signals were path is is completely completelydisconnected disconnectedbyby cuff-electrode path controller. obtained neural signals monitored and saved the devised GUI.devised The result showed the proposed wireless were monitored andusing saved using the GUI. Thethat result showed implantable that the proposed neural interface system couldinterface successfully record muscle afferent record activities during ramp-and-hold, implantable wireless neural system could successfully muscle afferent activities flexion-extension movements. during ramp-and-hold, flexion-extension movements. Full Extension

130 115 100 85 70

Extension

Flexion

Full Flexion 0

2

4

6

8

10

12

14

16

18

20

10

12

14

16

18

20

10

12

14

16

18

20

Tibial Nerve Signal [µV]

Peroneal Nerve Signal [µV]

Time [s] (a) 6 3 0 -3 -6 0

2

4

6

8

0

2

4

6

8

Time [s] (b)

6 3 0 -3 -6

Time [s] (c)

Figure signals during during the the passive passive movements movements of of the the rabbit’s rabbit’s ankle: ankle: Figure 17. 17. Acquired Acquired joint joint angle angle and and ENG ENG signals (a) joint angle, (b) ENG signals from the peroneal nerve and (c) ENG signals from the tibial nerve. (a) joint angle, (b) ENG signals from the peroneal nerve and (c) ENG signals from the tibial nerve.

3.5.2. and Stimulation Stimulation 3.5.2. Performance Performance of of Simultaneous Simultaneous Neural Neural Signal Signal Recording Recording and For with the the electrical electrical stimulation, stimulation, the For neural neural signal signal recording recording with the neural neural stimulator stimulator was was only only connected to the cuff electrode during electrical stimulation period by the cuff-electrode path connected to the cuff electrode during electrical stimulation period by the cuff-electrode path controller. controller. The obtained signals were saved using theshown devised GUI. As The obtained neural signalsneural were monitored and monitored saved usingand the devised GUI. As in Figure 18, shown in Figure 18, when electrical stimulation was applied to the peroneal nerve, the dorsi-flexor when electrical stimulation was applied to the peroneal nerve, the dorsi-flexor muscles were activated muscles and to thethe ankle was rotated to thestrong full-flexion strong ENG signals and the were ankleactivated was rotated full-flexion position; ENG position; signals were recorded from were recorded from the opposite tibial nerve when the ankle reached full-flexion. On other the opposite tibial nerve when the ankle reached full-flexion. On the other hand, when the electrical hand, whenwas electrical applied to the tibialmuscles nerve, the stimulation appliedstimulation to the tibialwas nerve, the plantar-flexor wereplantar-flexor activated andmuscles the anklewere was activated and the ankle was rotated to the full-extension position; strong ENG signals were rotated to the full-extension position; strong ENG signals were recorded from the opposite peroneal recorded from opposite peroneal nerve when the ankle reached full-extension. Otherwise (i.e., nerve when thethe ankle reached full-extension. Otherwise (i.e., when there was no stimulation), the when there was no stimulation), the ankle moved to the neutral position without showing ankle moved to the neutral position without showing dominant ENG signals from either the tibial or dominant ENG signals from either the tibial or peroneal nerve. As shown in Figure 18e, there was no undershoot, overshoot, or ringing and the rabbit’s ankle showed mild movements by the

Sensors 2018, 18, 1

19 of 26

Sensors 2018, 18, 1

19 of 26

250 125 0 -125 -250 0

1

2

3

4

5 Time [s] (a)

6

7

8

9

10

0

1

2

3

4

5 Time [s] (b)

6

7

8

9

10

0

1

2

3

4

5 Time [s] (c)

6

7

8

9

10

0

1

2

3

4

5 Time [s] (d)

6

7

8

9

10

4

5 Time [s] (e)

6

7

8

9

10

250 125 0 -125 -250

Tibial Nerve Signal [µV]

6 3 0 -3 -6

6 3 0 -3 -6

Ankle Angle [degree]

Peroneal Nerve Signal [µV]

Stimulation Waveform Stimulation Waveform for Peroneal Nerve [µA] for Tibial Nerve [µA]

peroneal nerve. As shown in Figure 18e, there was no undershoot, overshoot, or ringing and the rabbit’s delicate biphasic including ramp-up and waveforms ramp-downincluding periods that were ankle showed mildstimulation movementswaveforms by the delicate biphasic stimulation ramp-up intentionally designed to prevent sudden muscle contraction and relaxation. The devised and ramp-down periods that were intentionally designed to prevent sudden muscle contraction and implantable device worked as designed forworked concurrent neural signal recording and stimulation. The relaxation. The devised implantable device as designed for concurrent neural signal recording cuff-electrode path controller made path it possible to use oneit cuff electrode both and stimulation. The cuff-electrode controller made possible to usefor one cuffrecording electrode and for stimulation and thestimulation blanking circuit properly blocked interference stimulus artefacts, evoked both recording and and the blanking circuit the properly blockedofthe interference of stimulus ENG andevoked EMG signals caused bysignals the electrical artefacts, ENG and EMG causedstimulations. by the electrical stimulations.

130 115 100 85 70

Full Extension Full Flexion 0

1

2

3

Figure18. 18.Acquired Acquired joint angle signals the electrical stimulation period: (a) Figure joint angle andand ENGENG signals duringduring the electrical stimulation period: (a) biphasic biphasic stimulation the tibial nerve,stimulation (b) biphasic stimulation the stimulation waveformswaveforms for the tibialfor nerve, (b) biphasic waveforms for waveforms the peronealfor nerve, peroneal nerve,from (c) ENG signals from the(d) peroneal nerve,from (d) ENG signals from (c) ENG signals the peroneal nerve, ENG signals the tibial nerve andthe (e)tibial joint nerve angle. and (e) joint angle.

3.5.3. Performance of Blanking Circuit during Simultaneous Neural Signal Recording and Stimulation An additional experiment was conducted to evaluate the performance of the blanking circuit. Figure 19a shows the acquired neural signals including stimulation artifacts, evoked ENG signals,

Sensors 2018, 18, 1

20 of 26

3.5.3. Performance of Blanking Circuit during Simultaneous Neural Signal Recording and Stimulation Sensors 2018, 18, 1

20 of 26

Muscle Afferent Signals [µV]

Stimulation Artifacts, Evoked ENG/EMG & Muscle Afferent Signals [mV]

An additional experiment was conducted to evaluate the performance of the blanking circuit. evoked19a EMG signals muscle afferent signals. This data recording was performed withsignals, 40-dB Figure shows the and acquired neural signals including stimulation artifacts, evoked ENG gain without using the blanking circuit. The stimulation artifacts, evoked ENG and EMG signals evoked EMG signals and muscle afferent signals. This data recording was performed with 40-dB gain appeared in synchrony with 50-Hz stimulation, which interfered with signals the recording of without using the blanking circuit. Theelectrical stimulation artifacts, evoked ENG and EMG appeared muscle afferent signals. this issue could be interfered avoided by using devisedofblanking circuit. in synchrony with 50-HzHowever, electrical stimulation, which with thethe recording muscle afferent Figure 19b showsthis theissue obtained afferent signals theblanking blanking circuit was 19b applied to signals. However, couldmuscle be avoided by using the when devised circuit. Figure shows eliminate themuscle interference. datawhen recording was performed with 100-dB gain and the a synchronized the obtained afferentThis signals the blanking circuit was applied to eliminate interference. blanking withwas eachperformed stimulus switched the gain neural signal to ground for a duration 8 ms. By This data pulse recording with 100-dB and a synchronized blanking pulseofwith each using blanking circuit, the neural signals of interest could be recorded without the interference from stimulus switched the neural signal to ground for a duration of 8 ms. By using blanking circuit, the the electrical neural signalsstimulation. of interest could be recorded without the interference from the electrical stimulation.

0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 -0.5

Evoked ENG Muscle Afferent

Stimulation Artifact

Evoked EMG 0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.06

0.07

Time [s] (a) 9

Muscle Afferent

Blanking Period

6 3 0 -3 -6 -9

0

0.01

0.02

0.03

0.04

0.05

Time [s] (b)

Figure 19. 19. Acquired Acquired neural neural signals: signals: (a) (a) when when the the blanking blanking processes processes were were not not applied applied and and (b) (b) when when Figure the blanking processes were applied. the blanking processes were applied.

4. Discussion Discussion 4. In this this study, system, which cancan be In study, we we proposed proposed aanovel novelimplantable implantablewireless wirelessneural neuralinterface interface system, which rapidly implemented and modified with cost effective COTS components. The neural interface be rapidly implemented and modified with cost effective COTS components. The neural interface system includes includes aa WPC-compliant system WPC-compliant power power transmission transmissioncircuit, circuit,an anMICS-band-based MICS-band-basedradio radiolink linkand anda cuff-electrode path controller for simultaneous neural signal recording and stimulation using a cuff-electrode path controller for simultaneous neural signal recording and stimulation using aa single cuff cuff electrode. electrode. The The maximum maximum reliable reliable operating operating distance distance for for wireless wireless power power transmission transmission was was single found to tobe beapproximately approximately mm overall efficiency was measured to bewhich 67%, is which is found 1111 mm andand the the overall efficiency was measured to be 67%, higher higher than conventional wireless power transmission devices [7,9,10,48,49]. In addition, the than conventional wireless power transmission devices [7,9,10,48,49]. In addition, the wireless data wireless data communication guaranteed sufficient data rate 400 kbps while consuming low communication guaranteed a sufficient dataarate of 400 kbps whileofconsuming low power (receiving: power (receiving: 4.7 mA, transmitting: 5.6 mA) and providing a longer data transmission range of 4.7 mA, transmitting: 5.6 mA) and providing a longer data transmission range of 1.98 m compared to 1.98 m compared to devices that use inductive or optical communications [8,9,11,19,50]. The devices that use inductive or optical communications [8,9,11,19,50]. The cuff-electrode path controller cuff-electrode path controller made it possible use one cuff electrode at the samesignal implant site for made it possible to use one cuff electrode at thetosame implant site for both neural recording both neural signal recording and stimulation, enabling the cuff electrode path to be shared for the and stimulation, enabling the cuff electrode path to be shared for the neural amplifier and stimulator neural amplifier and stimulator based on its control signal. The devised system allowed the based on its control signal. The devised system allowed the experimenter to select a wide range experimenter to select a wide range of stimulation and recording parameters. Through the GUI software, the neural stimulation and recording parameters could be wirelessly set and the acquired neural signals could be displayed in real-time and archived using a bidirectional MICS-band-based radio link.

Sensors 2018, 18, 1

21 of 26

of stimulation and recording parameters. Through the GUI software, the neural stimulation and recording parameters could be wirelessly set and the acquired neural signals could be displayed in real-time and archived using a bidirectional MICS-band-based radio link. Table 8 provides a comparison of the system presented in this paper with other systems published in recent years, including various elements: the numbers of neural signal recording channel and stimulation channel, gain, sampling rate, data communication, power supply, target area, weight and size. It is noted that we have limited the comparison to the studies that implemented wireless neural interface devices using COTS components only. The functionality of the previous studies [9,12,49] is restricted while solely working as a neural signal recorder or a neural stimulator. Hence, extra devices are required for the more sophisticated experiments where concurrent neural stimulation and recording are required. The neural interface devices introduced in [51,52] are equipped with both functions of neural recording and neural stimulation by using separated modules. However, it is difficult to implant the separated modules with a large size into the limited inner space of a body. Moreover, individual power sources, power regulating components and control processors are needed for each module. Contrary to these, the proposed system was integrated into one compact module allowing a simple insertion procedure. One compact implantable module has advantage in the aspect of aesthetic since it occupies small area without noticing by other people. Therefore, one compact module is desirable for implantable devices. Extra surgeries should be refrained as much as possible but using limited power sources [9,12,49,51–54] such as non-rechargeable batteries cannot avoid extra surgeries to replace the exhausted batteries with new one. To prevent extra surgeries, inductive power transmission and ultrasonic power transmission technologies are being used for supplying power to the implanted devices without rechargeable batteries. However, these methods highly restrict patients’ movement because they require very close contact to operate. To solve these issues, extra surgeries and limited mobility, we combined two strategies, the WPT compliant inductive power transmission and the rechargeable battery. Although patients’ movement is temporary limited during charging procedure, the patients can freely move after the charging procedure. Also, there is no need for extra surgeries to change an exhausted battery. Establishing an implantable data communication is possible by other ISM bands [9,12,49,51–54] but they are not as qualified as the MICS band for implantable medical devices. Furthermore, they are prone to establish vulnerable data communication and consume higher power when compared to the MICS band. The MICS band consumes smaller power than other radio bands to communicate between the implantable device and the external device since the attenuation of the RF signal is smaller in comparison with other bands in the human body. The ZL70102 MICS band RF transceiver used in this study consumes about 5 mA, while the other RF transceivers in Table 8 consume higher current: TX3A (7.5 mA), TX2 (12 mA), nRF24L01 (12 mA) and CC2500 (13 mA). The MICS band using the frequency range at 402–405 MHz is exclusively allocated for implantable biotelemetry applications by the FCC. Therefore, unexpected interference can be avoided with specifically designed channel bandwidth, output power level and duty cycle. Using the MICS band allowed the devised system to establish a robust, secure and low-power consuming wireless data communication link.

Sensors 2018, 18, 1

22 of 26

Table 8. Comparison of wireless neural interface systems. This Work

Liang et al. [9]

Xu et al. [49]

Zhou et al. [12]

Ativanichayaphong et al. [53]

Liu et al. [51]

Pinnell et al. [54]

Zuo et al. [52]

Recording Channels

2

1

0

0

1

4

4

1

Stimulation Channels

2

0

2

1

1

6

2

1

40–120 dB

100 dB

-

-

84 dB

54–96 dB

34 dB

73 dB

10 kSPS

11 kSPS

-

-

10 kSPS

30 kSPS

500 SPS

10 kSPS

ZL70102 (MICS)

Inductive Link

CC2500 (ISM), Inductive Link

CC2500 (ISM)

TX3A, RX2A (ISM)

nRF24L01 (ISM)

CC2500 (ISM)

nRF24L01 (ISM)

WPT, Rechargeable battery

Inductive coupling

Inductive coupling

Battery

Battery

Battery

Battery

Battery

Tibial nerve, Peroneal nerve

Sciatic nerve

Lumbar-sacral segment

Lumber-sacral segment

Dorsal horn, Periaqueductal gray, Anterior cingulate cortex

Brain

Brain

Dorsal horn, Periaqueductal gray

21 g

-

3.78 g

12.6 g

20 g

-

8.5 g

12 g (STIM), 15 g (REC)

25 × 50 × 27 mm

56 × 36 × 15 mm (REC), 43 × 27 × 8 mm (STIM), 31 × 13 × 8 mm (BSN)

28 × 17 × 7 mm

27 × 8 mm (REC), 27 × 12 mm (STIM)

Gain Sampling Rate Data Communication (Band) Power Supply

Target Area

Weight

Size

33 × 28 × 12 mm

45 × 30 × 12 mm

22 × 23 × 7 mm

33 × 24 × 8 mm

Sensors 2018, 18, 1

23 of 26

5. Conclusions The results obtained during bench tests and animal experiments clearly showed that the devised implantable wireless neural interface system could simultaneously record neural activities and stimulate nerves while sharing a single cuff electrode. Furthermore, the devised system not only includes various functions: a neural amplifier, a neural stimulator, a cuff path controller, a blanking circuit, a wireless data transceiver, a wireless power receiver but also allows adjusting various recording and stimulation parameters: the gain of a neural amplifier, the stimulation periods, the stimulation amplitudes and the number of biphasic stimulation waveforms avoiding sudden muscle contraction and relaxation. Thus, the system proposed in this study is highly versatile and can be easily reconfigured for various implantable medical devices by their own purpose, especially such as closed-loop control based implantable neural prostheses requiring neural signal recording and stimulation at the same time. Although the animals used in this study were anesthetized to assess the developed system, further studies will be performed with freely moving animals by using the devised system for many potential applications in neuroscience requiring nerve stimulation and/or neural signal recording. In addition, more research on secure hermetic packaging and heating issues should be studied for long term implantation. Acknowledgments: This research was partially supported by the convergence technology development program for bionic arm through the National Research Foundation of Korea (NRF) funded by the Ministry of Science & ICT (2017M3C1B2085311), a grant of the Korea Health Technology R & D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & welfare, Republic of Korea (HI14C3477) and the Korea Institute of Science and Technology Institutional Program (2E6890). Author Contributions: Ahnsei Shon, Jun-Uk Chu and Inchan Youn conceived and designed the experiments; Ahnsei Shon, Jun-Uk Chu and Jiuk Jung designed hardware system, firmware and software; Ahnsei Shon and Hyungmin Kim performed the experiments; Ahnsei Shon, Jun-Uk Chu and Inchan Youn analyzed the data; Ahnsei Shon and Jun-Uk Chu wrote the paper. Conflicts of Interest: The authors declare no conflict of interest. Ethical Statements: The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Institutional Animal Care and Use Committee of the Korea Institute of Science and Technology (approval number: 2017-019).

References 1. 2.

3.

4.

5.

6. 7.

Haugland, M.K.; Sinkjaer, T. Cutaneous whole nerve recordings used for correction of footdrop in hemiplegic man. IEEE Trans. Rehabil. Eng. 1995, 3, 307–317. [CrossRef] Kyberd, P.J.; Holland, O.E.; Chappell, P.H.; Smith, S.; Tregidgo, R.; Bagwell, P.J.; Snaith, M. MARCUS: A two degree of freedom hand prosthesis with hierarchical grip control. IEEE Trans. Rehabil. Eng. 1995, 3, 70–76. [CrossRef] Saper, J.R.; Dodick, D.W.; Silberstein, S.D.; McCarville, S.; Sun, M.; Goadsby, P.J. Occipital nerve stimulation for the treatment of intractable chronic migraine headache: ONSTIM feasibility study. Cephalalgia 2011, 31, 271–285. [CrossRef] [PubMed] Boyer, S.; Sawan, M.; Abdel-Gawad, M.; Robin, S.; Elhilali, M.M. Implantable selective stimulator to improve bladder voiding: Design and chronic experiments in dogs. IEEE Trans. Rehabil. Eng. 2000, 8, 464–470. [CrossRef] [PubMed] Chu, J.-U.; Song, K.-I.; Han, S.; Lee, S.H.; Kim, J.; Kang, J.Y.; Hwang, D.; Suh, J.-K.F.; Choi, K.; Youn, I. Improvement of signal-to-interference ratio and signal-to-noise ratio in nerve cuff electrode systems. Physiol. Meas. 2012, 33, 943. [CrossRef] [PubMed] Perry, D.W.J.; Grayden, D.B.; Shepherd, R.K.; Fallon, J.B. A fully implantable rodent neural stimulator. J. Neural Eng. 2012, 9, 014001. [CrossRef] [PubMed] Yin, M.; Borton, D.A.; Aceros, J.; Patterson, W.R.; Nurmikko, A.V. A 100-Channel Hermetically Sealed Implantable Device for Chronic Wireless Neurosensing Applications. IEEE Trans. Biomed. Circuits Syst. 2013, 7, 115–128. [CrossRef] [PubMed]

Sensors 2018, 18, 1

8.

9.

10. 11. 12.

13. 14. 15.

16.

17.

18. 19. 20. 21.

22. 23. 24. 25.

26.

27. 28.

24 of 26

Borton, D.A.; Song, Y.K.; Patterson, W.R.; Bull, C.W.; Park, S.; Laiwalla, F.; Donoghue, J.P.; Nurmikko, A.V. Wireless, high-bandwidth recordings from non-human primate motor cortex using a scalable 16-Ch implantable microsystem. In Proceedings of the 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Minneapolis, MN, USA, 3–6 September 2009; pp. 5531–5534. Liang, C.-K.; Chen, J.-J.J.; Chung, C.-L.; Cheng, C.-L.; Wang, C.-C. An implantable bi-directional wireless transmission system for transcutaneous biological signal recording. Physiol. Meas. 2005, 26, 83. [CrossRef] [PubMed] Rizk, M.; Bossetti, C.A.; Jochum, T.A.; Callender, S.H.; Nicolelis, M.A.L.; Turner, D.A.; Wolf, P.D. A fully implantable 96-channel neural data acquisition system. J. Neural Eng. 2009, 6, 026002. [CrossRef] [PubMed] Sodagar, A.M.; Perlin, G.E.; Yao, Y.; Najafi, K.; Wise, K.D. An Implantable 64-Channel Wireless Microsystem for Single-Unit Neural Recording. IEEE J. Solid-State Circuits 2009, 44, 2591–2604. [CrossRef] Zhou, H.; Xu, Q.; He, J.; Ren, H.; Zhou, H.; Zheng, K. A fully implanted programmable stimulator based on wireless communication for epidural spinal cord stimulation in rats. J. Neurosci. Methods 2012, 204, 341–348. [CrossRef] [PubMed] Bogónez-Franco, P.; Nescolarde, L.; Gálvez-Montón, C.; Bragós, R.; Rosell-Ferrer, J. An implantable bioimpedance monitor using 2.45 GHz band for telemetry. Physiol. Meas. 2013, 34, 1. [CrossRef] [PubMed] Clarke, C.T.; Xu, X.; Rieger, R.; Taylor, J.; Donaldson, N. An implanted system for multi-site nerve cuff-based ENG recording using velocity selectivity. Analog Integr. Circuits Signal Process. 2009, 58, 91–104. [CrossRef] Guiraud, D.; Stieglitz, T.; Koch, K.P.; Divoux, J.-L.; Rabischong, P. An implantable neuroprosthesis for standing and walking in paraplegia: 5-Year patient follow-up. J. Neural Eng. 2006, 3, 268. [CrossRef] [PubMed] Kong, W.; Huang, J.; Rollins, D.L.; Ideker, R.E.; Smith, W.M. A semi-implantable multichannel telemetry system for continuous electrical, mechanical and hemodynamical recordings in animal cardiac research. Physiol. Meas. 2007, 28, 249. [CrossRef] [PubMed] Lee, S.B.; Lee, H.M.; Kiani, M.; Jow, U.M.; Ghovanloo, M. An Inductively Powered Scalable 32-Channel Wireless Neural Recording System-on-a-Chip for Neuroscience Applications. IEEE Trans. Biomed. Circuits Syst. 2010, 4, 360–371. [CrossRef] Rush, A.D.; Troyk, P.R. A Power and Data Link for a Wireless-Implanted Neural Recording System. IEEE Trans. Biomed. Eng. 2012, 59, 3255–3262. [CrossRef] [PubMed] Jegadeesan, R.; Nag, S.; Agarwal, K.; Thakor, N.V.; Guo, Y.X. Enabling Wireless Powering and Telemetry for Peripheral Nerve Implants. IEEE J. Biomed. Health Inform. 2015, 19, 958–970. [CrossRef] [PubMed] Peng, C.-W.; Chen, J.-J.J.; Lin, C.-C.K.; Poon, P.W.-F.; Liang, C.-K.; Lin, K.-P. High frequency block of selected axons using an implantable microstimulator. J. Neurosci. Methods 2004, 134, 81–90. [CrossRef] [PubMed] Valdastri, P.; Rossi, S.; Menciassi, A.; Lionetti, V.; Bernini, F.; Recchia, F.A.; Dario, P. An implantable ZigBee ready telemetric platform for in vivo monitoring of physiological parameters. Sens. Actuators Phys. 2008, 142, 369–378. [CrossRef] Islam, M.N.; Yuce, M.R. Review of Medical Implant Communication System (MICS) band and network. ICT Express 2016, 2, 188–194. [CrossRef] Fang, Q.; Lee, S.Y.; Permana, H.; Ghorbani, K.; Cosic, I. Developing a Wireless Implantable Body Sensor Network in MICS Band. IEEE Trans. Inf. Technol. Biomed. 2011, 15, 567–576. [CrossRef] [PubMed] Haugland, M.K.; Hoffer, J.A. Artifact-free sensory nerve signals obtained from cuff electrodes during functional electrical stimulation of nearby muscles. IEEE Trans. Rehabil. Eng. 1994, 2, 37–40. [CrossRef] Jensen, W.; Sinkjaer, T.; Sepulveda, F. Improving signal reliability for on-line joint angle estimation from nerve cuff recordings of muscle afferents. IEEE Trans. Neural Syst. Rehabil. Eng. 2002, 10, 133–139. [CrossRef] [PubMed] Micera, S.; Jensen, W.; Sepulveda, F.; Riso, R.R.; Sinkjaer, T. Neuro-fuzzy extraction of angular information from muscle afferents for ankle control during standing in paraplegic subjects: An animal model. IEEE Trans. Biomed. Eng. 2001, 48, 787–794. [CrossRef] [PubMed] Nielsen, T.N.; Sevcencu, C.; Struijk, J.J. Fascicle-Selectivity of an Intraneural Stimulation Electrode in the Rabbit Sciatic Nerve. IEEE Trans. Biomed. Eng. 2012, 59, 192–197. [CrossRef] [PubMed] Strange, K.D.; Hoffer, J.A. Restoration of use of paralyzed limb muscles using sensory nerve signals for state control of FES-assisted walking. IEEE Trans. Rehabil. Eng. 1999, 7, 289–300. [CrossRef] [PubMed]

Sensors 2018, 18, 1

29. 30. 31.

32. 33. 34.

35. 36. 37. 38.

39.

40.

41.

42. 43. 44. 45. 46.

47.

48. 49.

25 of 26

Triantis, I.F.; Demosthenous, A.; Donaldson, N. On cuff imbalance and tripolar ENG amplifier configurations. IEEE Trans. Biomed. Eng. 2005, 52, 314–320. [CrossRef] [PubMed] Gaxiola-Sosa, J.E.; Entesari, K. Design and implementation of a wireless medical system prototype for implantable applications. Analog Integr. Circuits Signal Process. 2015, 82, 99–111. [CrossRef] Lee, S.H.; Jung, J.H.; Chae, Y.M.; Suh, J.-K.F.; Kang, J.Y. Fabrication and characterization of implantable and flexible nerve cuff electrodes with Pt, Ir and IrO x films deposited by RF sputtering. J. Micromech. Microeng. 2010, 20, 035015. [CrossRef] Nikolic, Z.M.; Popovic, D.B.; Stein, R.B.; Kenwell, Z. Instrumentation for ENG and EMG recordings in FES systems. IEEE Trans. Biomed. Eng. 1994, 41, 703–706. [CrossRef] [PubMed] Bertemes-Filho, P.; Brown, B.H.; Wilson, A.J. A comparison of modified Howland circuits as current generators with current mirror type circuits. Physiol. Meas. 2000, 21, 1. [CrossRef] [PubMed] Chu, J.-U.; Song, K.-I.; Shon, A.; Han, S.; Lee, S.H.; Kang, J.Y.; Hwang, D.; Suh, J.-K.F.; Choi, K.; Youn, I. Feedback control of electrode offset voltage during functional electrical stimulation. J. Neurosci. Methods 2013, 218, 55–71. [CrossRef] [PubMed] Bhadra, N.; Kilgore, K.L.; Peckham, P.H. Implanted stimulators for restoration of function in spinal cord injury. Med. Eng. Phys. 2001, 23, 19–28. [CrossRef] Gwilliam, J.C.; Horch, K. A charge-balanced pulse generator for nerve stimulation applications. J. Neurosci. Methods 2008, 168, 146–150. [CrossRef] [PubMed] Nag, S.; Thakor, N.V. Implantable neurotechnologies: Electrical stimulation and applications. Med. Biol. Eng. Comput. 2016, 54, 63–76. [CrossRef] [PubMed] Byrne, C.A.; O’Keeffe, D.T.; Donnelly, A.E.; Lyons, G.M. Effect of walking speed changes on tibialis anterior EMG during healthy gait for FES envelope design in drop foot correction. J. Electromyogr. Kinesiol. 2007, 17, 605–616. [CrossRef] [PubMed] Bradley, P.D. Wireless medical implant technology—Recent advances and future developments. In Proceedings of the 2011 Proceedings of the European Solid-State Device Research Conference (ESSDERC), Helsinki, Finland, 12–16 September 2011; pp. 54–58. Seckler, T.; Jagielski, K.; Stunder, D. Assessment of Electromagnetic Interference with Active Cardiovascular Implantable Electronic Devices (CIEDs) Caused by the Qi A13 Design Wireless Charging Board. Int. J. Environ. Res. Public. Health 2015, 12, 5886–5904. [CrossRef] [PubMed] An Introduction to the Wireless Power Consortium Standard and TI’s Compliant Solutions. Available online: https://www.mouser.com/pdfDocs/An-introduction-to-the-Wireless-Power-Consortium.pdf (accessed on 24 November 2017). Designing a Qi-Compliant Receiver Coil for Wireless Power Systems, Part 1. Available online: https://www. mouser.com/pdfDocs/TI-Designing-a-Qi-compliant-receiver-coil.pdf (accessed on 24 November 2017). Hui, S.Y. Planar Wireless Charging Technology for Portable Electronic Products and Qi. Proc. IEEE 2013, 101, 1290–1301. [CrossRef] Lin, C.-C.K.; Ju, M.-S.; Cheng, H.-S. Model-based ankle joint angle tracing by cuff electrode recordings of peroneal and tibial nerves. Med. Biol. Eng. Comput. 2007, 45, 375–385. [CrossRef] [PubMed] Jensen, W.; Lawrence, S.M.; Riso, R.R.; Sinkjaer, T. Effect of initial joint position on nerve-cuff recordings of muscle afferents in rabbits. IEEE Trans. Neural Syst. Rehabil. Eng. 2001, 9, 265–273. [CrossRef] [PubMed] Riso, R.R.; Mosallaie, F.K.; Jensen, W.; Sinkjaer, T. Nerve cuff recordings of muscle afferent activity from tibial and peroneal nerves in rabbit during passive ankle motion. IEEE Trans. Rehabil. Eng. 2000, 8, 244–258. [CrossRef] [PubMed] Donaldson, N.D.N.; Zhou, L.; Perkins, T.A.; Munih, M.; Haugland, M.; Sinkjaer, T. Implantable telemeter for long-term electroneurographic recordings in animals and humans. Med. Biol. Eng. Comput. 2003, 41, 654–664. [CrossRef] Su, Y.; Routhu, S.; Moon, K.S.; Lee, S.Q.; Youm, W.; Ozturk, Y. A Wireless 32-Channel Implantable Bidirectional Brain Machine Interface. Sensors 2016, 16. [CrossRef] [PubMed] Xu, Q.; Hu, D.; Duan, B.; He, J. A Fully Implantable Stimulator With Wireless Power and Data Transmission for Experimental Investigation of Epidural Spinal Cord Stimulation. IEEE Trans. Neural Syst. Rehabil. Eng. 2015, 23, 683–692. [CrossRef] [PubMed]

Sensors 2018, 18, 1

50.

51.

52. 53.

54.

26 of 26

Liu, T.; Bihr, U.; Anis, S.M.; Ortmanns, M. Optical transcutaneous link for low power, high data rate telemetry. In Proceedings of the 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, San Diego, CA, USA, 28 August–1 September 2012; pp. 3535–3538. Liu, X.; Subei, B.; Zhang, M.; Richardson, A.G.; Lucas, T.H.; van Der Spiegel, J. The PennBMBI: Design of a general purpose wireless brain-machine-brain interface system. IEEE Trans. Biomed. Circuits Syst. 2015, 9, 248–258. [CrossRef] [PubMed] Zuo, C.; Yang, X.; Wang, Y.; Hagains, C.E.; Li, A.-L.; Peng, Y.B.; Chiao, J.-C. A digital wireless system for closed-loop inhibition of nociceptive signals. J. Neural Eng. 2012, 9, 056010. [CrossRef] [PubMed] Ativanichayaphong, T.; He, J.W.; Hagains, C.E.; Peng, Y.B.; Chiao, J.C. A combined wireless neural stimulating and recording system for study of pain processing. J. Neurosci. Methods 2008, 170, 25–34. [CrossRef] [PubMed] Pinnell, R.C.; Dempster, J.; Pratt, J. Miniature wireless recording and stimulation system for rodent behavioural testing. J. Neural Eng. 2015, 12, 066015. [CrossRef] [PubMed] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).