Floating EMG sensors and stimulators wirelessly powered and operated by volume conduction for networked neuroprosthetics

Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 https://doi.org/10.1186/s12984‑022‑01033‑3 RESEARCH Floating EMG sensors andstimulators wirelessly powered andoperated byvolume conduction fornetworked neuroprosthetics Laura Becerra‑Fajardo1*, Marc Oliver Krob2, Jesus Minguillon1,3,4, Camila Rodrigues5,6, Christine Welsch2, Marc Tudela‑Pi1, Albert Comerma1, Filipe Oliveira Barroso5, Andreas Schneider2 and Antoni Ivorra1,7 Abstract Background: Implantable neuroprostheses consisting of a central electronic unit wired to electrodes benefit thou‑ sands of patients worldwide. However, they present limitations that restrict their use. Those limitations, which are more adverse in motor neuroprostheses, mostly arise from their bulkiness and the need to perform complex surgical implantation procedures. Alternatively, it has been proposed the development of distributed networks of intramus‑ cular wireless microsensors and microstimulators that communicate with external systems for analyzing neuromus‑ cular activity and performing stimulation or controlling external devices. This paradigm requires the development of miniaturized implants that can be wirelessly powered and operated by an external system. To accomplish this, we propose a wireless power transfer (WPT) and communications approach based on volume conduction of innocuous high frequency (HF) current bursts. The currents are applied through external textile electrodes and are collected by the wireless devices through two electrodes for powering and bidirectional digital communications. As these devices do not require bulky components for obtaining power, they may have a flexible threadlike conformation, facilitating deep implantation by injection. Methods: We report the design and evaluation of advanced prototypes based on the above approach. The system consists of an external unit, floating semi‑implantable devices for sensing and stimulation, and a bidirectional com‑ munications protocol. The devices are intended for their future use in acute human trials to demonstrate the distrib‑ uted paradigm. The technology is assayed in vitro using an agar phantom, and in vivo in hindlimbs of anesthetized rabbits. Results: The semi‑implantable devices were able to power and bidirectionally communicate with the external unit. Using 13 commands modulated in innocuous 3 MHz HF current bursts, the external unit configured the sensing and stimulation parameters, and controlled their execution. Raw EMG was successfully acquired by the wireless devices at 1 ksps. Conclusions: The demonstrated approach overcomes key limitations of existing neuroprostheses, paving the way to the development of distributed flexible threadlike sensors and stimulators. To the best of our knowledge, these devices are the first based on WPT by volume conduction that can work as EMG sensors and as electrical stimulators in a network of wireless devices. © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Open Access *Correspondence: [email protected] 1 Department of Information and Communications Technologies, Universitat Pompeu Fabra, 08018 Barcelona, Spain Full list of author information is available at the end of the article Page 2 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 Background Clinically available neuroprostheses have been demonstrated to improve the quality of life of patients with neurological disorders or injuries, as these individuals gain significant functional improvement [1]. Additionally, some neuroprosthetic technologies have granted the possibility to perform intuitive control of robotic prostheses as they create an interface between the machine and the patient’s nervous system [2]. In the framework of the EXTEND collaborative project, funded by the European Commission and in which the authors participate, it has been defined the concept of Bidirectional Hyper-Connected Neural Systems (BHNS). The BHNS concept refers to systems consisting of minimally invasive communication links between multiple nerves or muscles in the body and external devices which may be interconnected between them (Fig.1). The concept requires the deployment of dense networks of wireless active implantable medical devices (AIMDs) that can perform distributed electrical stimulation and sensing. These wireless AIMDs must communicate in real time with external devices and tools that process and analyze the neuromuscular activity and control the stimulation and the action of machines. BHNS could, for example, be used for (1) tremor management in essential tremor and Parkinson’s disease by tremor prediction using electromyography (EMG) [3], and subsequent intramuscular stimulation [4]; (2) performing functional neuromuscular stimulation [5]; or (3) interfacing with assistive wearable robots for spinal cord injury (SCI) and interfacing with prostheses providing control and artificial sensory feedback [6]. Most neuroprostheses for chronic use, which are required to accomplish the BHNS concept, are implantable systems that consist of a central unit electrically connected to electrodes at target sites (e.g., cuff electrodes on peripheral nerves) by means of leads (i.e., wires) [7, 8]. However, these neuroprostheses present limitations that restrict their use. Those limitations, which are most detrimental in the case of motor neuroprostheses, often arise from their bulkiness and the need to perform complex surgical implantation procedures because of the leads [1, 9]. The leads also tend to fail due to mechanical stress, particularly in the case of motor neuroprostheses because electrodes are commonly deployed in long and mobile anatomical regions. As an alternative to these AIMDs based on central units, for the case of motor neuroprostheses, it was proposed and assayed the development of distributed networks of wireless intramuscular microstimulators that integrate the electronics and the electrodes [10, 11]. The devices communicate with external systems to perform neuromuscular stimulation aiming at motor restoration in patients suffering from motor paralysis [12]. The microstimulators were percutaneously implanted via injection, thereby avoiding complex surgeries. However, the devices were stiff and considerably large (diameters > 2mm), making them unsuitable for their use in a dense network of wireless microstimulators. The form factor of these wireless devices is mostly limited by the method used to power them. Batteries, with their intrinsic limited lifespan and large volume [13], are not suitable as primary sources of power in very small electronic implants. These floating implants, in clinical use or preclinically demonstrated, are typically powered Keywords: Wireless power transfer, Volume conduction, AIMDs, Bidirectional communications, Electromyography, Sensor, Electrical stimulation Fig. 1 Schematic representation of a Bidirectional Hyper‑Connected Neural System—BHNS. High frequency (> 1 MHz) volume conduction is used to wirelessly power and communicate with implantable devices for sensing and stimulation. The high‑level controller of the external system communicates with several low‑level control units that deliver current bursts through external textile electrodes. The envisioned wireless implantable devices will have a thread‑like conformation to facilitate their deployment by injection. These devices will consist of a flexible body with two electrodes at their opposite ends (Electrodes ‘A’ and ‘B’) and their electronics will be integrated in an ASIC Page 3 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 by means of wireless power transfer (WPT) methods such as inductive coupling—the most established WPT method—[14–17], ultrasonic acoustic coupling [18–20] and capacitive coupling [21–23]. These methods have obtained high miniaturization levels, at the expense of link efficiency, penetration depth, or functionality. Recent reviews on these methods can be found in [13, 24–26]. In the last years we have proposed and demonstrated very thin microstimulators whose operation is based on rectification of volume conducted innocuous high frequency (HF) current bursts to cause local low frequency currents capable of stimulation [27, 28]. A portion of the HF current picked up by the devices is not directly rectified to perform stimulation but used to power the electronics of the implant (e.g., for control and communications) and, in this sense, it can be understood that the implants employ WPT based on volume conduction. In fact, in a series of recent works, we have advocated for, and studied, the use of volume conducted HF current bursts applied through textile electrodes to power elongated implants in general, not only stimulators [29–31]. Remarkably, according to the approach we propose, the implants can be conceived as thin, flexible and elongated devices suitable for implantation by means of injection [27]. The approach uses the body as an electrical conductor of innocuous and imperceptible HF current bursts that are applied by an external system using external textile electrodes. The implants do not require bulky components within their body to be electrically fed, allowing the integration of the electronics in an application-specific integrated circuit (ASIC). They pick up a small portion of the energy available through their two electrodes located at the opposite ends of the implant. The energy is used to power and bidirectionally communicate with the external system (Fig.1). The elongated and ultrathin form factor allows the deployment of multiple implants in a single body region, favoring their use in dense networks of wireless AIMDs, such as those required by BHNS. In the general architecture of BHNS, the external system will consist of one top-level controller that communicates with the wireless AIMDs through several external low-level control units that act as bidirectional gateways (i.e., protocol translators) between the implants and the external controller (Fig. 1). The low-level units apply bursts of HF currents to power and communicate with the wireless devices. In Becerra-Fajardo etal. [27] we reported the development and invivo evaluation of semi-rigid, thin and addressable stimulators based on this WPT approach. The injectable devices were made only of off-the-shelf components, and had an overall diameter of 2 mm. Although successfully demonstrated in animals, due to their invasiveness, these devices are not adequate to conduct assays in humans. Therefore, to be able to acutely demonstrate in humans the feasibility of the BHNS concept, within the framework of the EXTEND project it was decided to develop semi-implantable devices that consist of ultrathin and short intramuscular electrodes, made with thin-film technology, that are connected to a miniature electronic circuit to be fixed on the skin. Not only the electronics of the devices were upgraded for EMG sensing [32] but they were also upgraded for communication capabilities. Here we report the development and evaluation of these semi-implantable devices and the external system that powers and controls them. To the best of our knowledge, the wireless devices presented here are the only devices based on WPT by volume conduction that can work both as EMG sensors and as electrical stimulators to form networks of wireless devices. Methods Developed system The two fundamental parts of the developed system are (1) the external system that delivers the HF current bursts for wireless powering and bidirectional communications from the external system to the floating semi-implantable devices (i.e., downlink), and from these devices to the external system (i.e., uplink), and (2) the floating devices. These floating devices are semi-implantable devices composed of intramuscular electrodes connected to a miniature external electronic circuit (Fig. 2). The external system, through the two external textile electrodes of a specific low-level control unit, delivers HF current bursts that power the floating devices located between the Fig. 2 Schematic representation of the floating semi‑implantable devices used for acute implantation, and a low‑level control unit of the external system. The semi‑implantable devices include injectable intramuscular electrodes and a miniature electronic circuit adhered to the skin. The low‑level unit of the external system delivers HF current bursts for powering, EMG sensing and electrical stimulation Page 4 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 textile electrodes. Using the same HF current bursts, the control unit sends commands to the wireless devices to operate them, and, depending on the command, delivers HF currents that are modulated by the wireless devices to generate an uplink reply. External system Hardware The external system integrates a low-power PC/104 single board computer (CMA34CRQ2100HR by RTD Embedded Technologies, Inc) acting as a high-level controller that can communicate with several low-level units. The basic architecture of the low-level control units is shown in Fig.3. The units include a small single-board computer—hereinafter “digital unit”—(Raspberry Pi 4 2G Model B, by Raspberry Pi Foundation) that controls the delivery of the HF current bursts and the bidirectional communications. They also include a HF generator and modulator (4064 by B&K Precision, Corp.) connected to a custom-made power amplifier consisting of five high voltage amplification modules, based on a high speed, high voltage operational amplifier (ADA4870 by Analog Devices, Inc.), which are connected in series through transformers. The output of the power amplifier is connected to a sensing resistor, and to a pair of textile electrodes attached to the skin. The sensing resistor, acting as a shunt resistor, is used by a custom-made demodulator for measuring the current flowing through the tissues. This current is monitored by the digital unit, especially during uplink, when the floating devices modulate their current consumption, creating minute variations in the current. Figure 3 shows a schematic representation of the modulating signal and resulting modulated signal for downlink and power bursts (blue waveforms), the voltage obtained across the sensing resistor, which corresponds to a downlink and an uplink section, and the corresponding demodulated uplink data (red waveforms). For illustration purposes, the uplink modulation index used in the scheme is 10%, higher than the < 1% modulation index of the load modulation performed by the floating devices. The load modulation scheme is explained below. For experimental convenience, in the present study the reported assays were performed without using the PC/104 single board computer as the top control device: the Raspberry Pi 4 acting as the digital unit was commanded through the Ethernet interface from a computer running Matlab (R2019b, by Mathworks, Inc). This computer acted as the high-level controller. Through its universal asynchronous receiver-transmitter (UART) port, the digital unit of the low-level unit (i.e., the Raspberry Pi 4) generates an amplitudeshift keying (ASK) signal to modulate the HF current bursts for downlink communications and for powering the wireless devices. For uplink communications, the digital unit reads the information amplified and filtered by the demodulator using the UART interface. For both downlink and uplink, the information is sent at a rate of 256kbps. This implies that each byte has a duration of 39.06µs (1 start bit + 1 byte + 1 stop bit). HF bursts for power The external system delivers an initial long HF sinusoidal burst coined “Power up”. This 30ms burst is required to power up the wireless devices located between the two external electrodes. After this, the wireless devices are kept energized, and running in an idle mode, by delivering short bursts with a repetition frequency (F) of 50Hz, and a duration (B) of 1.6ms. As later explained, the EMG analog front-end (AFE) of the wireless devices saturates during the burst. The repetition frequency and duration of these bursts is selected to (1) minimize their impact in the EMG front-end, favoring faster recovery from saturation and longer windows inbetween bursts that can be used for EMG acquisition; (2) obtain a duty cycle (D) that is low enough to avoid tissue heating, and (3) deliver enough energy to keep the semiimplantable devices powered. Digital communications A communication protocol stack structured in layers and based on the Open System Interconnection (OSI) model, has been created for performing the bidirectional communications between the external system and the wireless devices. The protocol was Fig. 3 Basic architecture of the low‑level control unit. The unit is governed by the high‑level controller and communicates with the wireless devices located in the tissue between the two external electrodes of the control unit. The device includes a digital unit that generates a modulating signal for a HF generator and modulator. The modulated HF current—carrying downlink information and bursts of power—is amplified and applied to the tissues through the external textile electrodes (blue waveforms). This same current is continuously monitored by the digital unit through a sensing resistor and a demodulator, to detect and decode minute current variations (< 1%) generated when the wireless devices do uplink (red waveforms) Page 5 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 created to ensure that the minimum required data frames were used in the downlink and uplink, minimizing the time the HF bursts are active. The active time of the HF bursts is related with the specific absorption rate (SAR), which is limited by safety standards (explained below). The application layer implemented in this protocol stack includes 13 different downlink commands to control (e.g., configure EMG acquisition and stimulation) and interrogate (e.g., ping, get samples) the wireless devices. The frames that encapsulate the commands and the replies used in the bidirectional communications are encoded employing Manchester coding, which offers two advantages: has no dc component to avoid charge injection when applying the HF current bursts, and it provides a first error detection mechanism. Other additional error detection mechanisms are parity bit, frame length and command code. A detailed description of the process performed by the external unit to command the wireless devices, request replies, and the timings used to deliver the HF current bursts is included in Additional file1: Methods. All downlink commands and uplink replies are preceded by one initialization byte. That is, each frame consists of one initialization byte followed by one or more bytes containing the information. All uplink replies are preceded by a downlink command. Between one downlink command (e.g., Get sample) and the corresponding uplink reply (e.g., Send sample) there is a period of 2.3ms in which no bursts are delivered by the external system. This time is required for processing purposes inside the wireless device. Additional file1: TableS1, reports the downlink commands included in the communication protocol stack, as well as the estimated transmission time required for them. The protocol allows to control up to 256 wireless devices located between two external electrodes. To avoid replicating the instructions to several wireless devices, the protocol enables the use of 256 groups of devices that can be configured simultaneously, or that can be requested to do a specific function (e.g., start sensing or stop sensing). This minimizes the delivery of HF bursts for downlink and avoids misusing the powering/communication channel. The stimulation and sensing configuration payloads of the communication protocol stack are reported in Additional file1: TableS2. Additional file1: TableS3, reports the uplink replies, description and timings used by the wireless devices to send information to the external system. The replies either correspond to (1) an acknowledge (ACK), (2) a sample or (3) the configuration information currently defined in the wireless device. When the external system sends a “Get configuration” request, the uplink frame is constructed using the same format as in the downlink. In the case of “Get sample” and “Retry sample”, the device replies with a frame consisting of 10 bits corresponding to the sample, and 2 more bits corresponding to an internal counter, which is used to control the correct uplink of samples. Compliance withelectrical safety standards ICNIRP and IEEE standards define safety levels with respect to human exposure to electromagnetic fields. These standards protect against health effects related to tissue heating [33, 34]. Heating limits are expressed by the standards in terms of the SAR, which can be related to the electric field at a point as: where σ is the conductivity of the tissue (S/m), ρ is the tissue density (kg/m3), and E is the electric field strength in tissue (V/m) averaged over 6min for local exposure. By applying the HF sinusoidal currents in the form of short bursts, the applied Erms is: where Epeak is the applied peak electric field (V/m), and D is the duty cycle. In the case of periodic bursts (frequency F; duration B), as is the case of the power maintenance bursts, D is equal to F·B. Equation(2) allows to calculate the applied Erms and SAR in terms of the Power up time, and the types of functions requested by the external system, including the commands sent in the downlink, and the information replied in the uplink. Additional file1: Tables S1, S3, can be used to calculate the duty cycle corresponding to the commands used in a bidirectional sequence. The duty cycle can be calculated as: where tPower up is the Power up time (i.e., 30ms), tact_comm corresponds to the active time of the commands, both downlink and uplink, tact_maint corresponds to the active time of the power maintenance bursts (which have a duty cycle equal to F·B), and taveraging is the averaging time of the rms (360s, equivalent to the 6min averaging time established for the SAR calculation). The duration of the power maintenance bursts is equivalent to taveraging minus the total time used for power up and bidirectional commands. (1) SAR = σ|Erms| 2 ρ (2) E rms = Epeak √ D √2 (3) D = t Power up +t act_comm +t act_maint t averaging Page 6 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 In the case of human limbs, the standards define a maximum SAR of 20 W/kg for persons in restricted environments at frequencies below 6 GHz, averaged over 10g cube of tissue [33, 34]. Wireless devices forEMG sensing andelectrical stimulation The wireless devices are composed of thin and flexible intramuscular electrodes made in thin-film technology, a miniature electronic circuit with off-the-shelf components, and an intermediate printed circuit board (PCB) that connects both parts. Intramuscular electrodes Active sites design: The intramuscular electrodes are based on similar thin-film electrodes reported in [35], which are intended for acute implantation. To define the geometry of the electrode contacts (i.e., the active sites, the actual electrodes strictly speaking), simulations were performed to test the ability of the electrodes to (1) obtain enough power to supply the wireless circuit for continuous EMG recording, and (2) generate stimulation pulses with amplitudes above 2mA and below 4mA. Other functions performed by the circuit were not evaluated as they require less power than that needed during EMG acquisition and electrical stimulation. The simulations were performed by modeling the tissues surrounding the intramuscular electrodes and the presence of the electric field applied by the external system using a Thévenin equivalent circuit [29]. See Additional file1: Methods, for details of the numerical methods for the intramuscular electrodes design. Besides power supply and current delivery capabilities, another aspect that was considered in the geometrical design of the active sites was the charge injection limit. If the applied pulses exceed the maximum charge injection capacity of the electrodes, irreversible reactions may occur, which can cause damage both to the electrodes and to the tissues. In the case of smooth platinum, the charge injection capacity is defined in the range between 50 and 150µC/cm2 for 200µs, biphasic, charge-balanced pulses [36]. In the case of microrough platinum, the charge injection capacity increases to 500µC/cm2 [37]. Final conformation: The final geometry of the electrodes is based on the results obtained with finite element method (FEM) simulations and with circuit simulations performed with SPICE. Both are described in Additional file1: Methods. The polyimide filament that contains the electrode contacts and the tracks has a width of 0.42mm, a length of 81.6mm, and a thickness of 0.02mm (Fig.4a). Between the centers of the contacts there is a distance of 30mm, and the contacts have a length of 7.5mm and a width of 0.265mm (Fig.4b). The edges of the contacts are rounded to avoid sharp corners that would lead to high current densities. The four electrode contacts are located on both faces of the filament: 2 contacts in the top layer, and 2 contacts in the bottom layer. The distal contacts (top and bottom) are electrically connected in the miniature electronic circuit, to use them as a single distal electrode with a total surface area of 3.8 mm2. This electrode, coined Electrode ‘A’ (Fig.4a), is the stimulation electrode when the semi-implantable device is in stimulation mode. The two proximal contacts (top and bottom) are also short-circuited, to use them as a single proximal electrode (Electrode ‘B’, return electrode in stimulation mode) with a total surface area of 3.8 mm2. Electrodes ‘A’ and ‘B’ are used as the inputs for the EMG analog frontend (AFE) and are indirectly used by the load modulator during uplink. Both circuits are explained below. Figure4c shows the construction of the intramuscular electrodes. In essence it consists of a polyimide substrate (thickness: 10µm) with a patterned metallization on both sides and a thin polyimide coating (thickness: 5µm) on both sides with openings for the active sites and the bonding pads. The metallization is uniform, that is, it is the same metallization both for the active sites and for the connection tracks from these to the bonding pads. A relatively thick layer of gold (450 nm) beneath the Fig. 4 Intramuscular electrodes. a Top view of the electrodes showing a distal contact (Electrode ‘A’), a proximal contact (Electrode ‘B’), and the bonding pads of the four electrode contacts. b Detail of the area where the distal and proximal electrodes are located. c Cross section of the intramuscular electrodes (not to scale). Gold and platinum are used as metallization layers for the electrode tracks and contacts (600 nm thick per layer), and polyimide is used as substrate and cover material. d Impedance measurement of the intramuscular electrodes in a bipolar arrangement (short circuited distal contacts against short circuited proximal contacts) in a 0.9% NaCl solution Page 7 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 platinum layer (150nm) wires ensures high conductivity for the tracks. Microrough platinum coating is used to increase the surface area of the electrodes, therefore increasing their charge injection capacity and improving power transfer. As the proximal end of the thin-film electrodes is connected to the external miniature electronic circuit, they have the same implantation limitation as those used in [35, 38]: the intramuscular electrodes cannot be implanted directly using a hypodermic needle as introducer, because the circuit would prevent the complete extraction of the needle after insertion. To overcome this drawback, the distal end of the filament containing the four electrode contacts has a U-shape structure with a guiding filament (Fig.4b) that facilitates the insertion of the electrodes in the muscle. The main filament with the active sites runs externally to the needle, while the thin guiding filament is inserted through the lumen of a 23G hypodermic needle (Sterican 4665600 by B. Braun Melsungen AG) having a length of 60mm and an outer diameter of 0.6mm. The end of the guiding filament is adhered to the Luer lock of the needle. To reduce the risk that the needle bevel cuts the guiding filament during implantation, the bevel is smoothed with a laser (Picco Laser, by O.R. Lasertechnologie, Germany) prior to inserting the guiding filament inside the needle. The small intermediate PCB is used for (1) doing the electroplating process to coat the electrode contacts with microrough platinum [39], and (2) connecting the injectable electrodes to the miniature electronic circuit. The polyimide electrodes are bonded to the PCB using the Microflex technology to achieve an electrical and mechanical connection [40]. After electroplating, the circuit of the semi-implantable device is stacked on top of the small PCB via surface-mount vertical headers, and the final electronic assembly is protected within a 3D printed housing (Fig.5). Miniature electronic circuit The electronic architecture of the wireless device used for sensing and stimulation and which is powered and operated by HF volume conduction is partially based on [27]. Yet the architecture presented here has several advantages over the design reported in the past, including better power efficiency, bidirectional communications with higher data rates, the integration of an AFE for EMG acquisition, an ultra-low power microcontroller with several peripherals and running a finite-state machine, and a communication protocol stack (described above) for a more robust control from the external system. The electronic circuit has a length of 10mm, and a width of 8 mm. The surface-mount vertical headers shown in Fig.5a, connect the circuit to the intermediate PCB that includes the intramuscular electrodes. The final conformation of the wireless device, including the insertion needle and the capsule to protect the wireless circuit, is shown in Fig.5b. Power regulation and dc-blocking: Figure 5e shows the basic architecture of the electronic circuit. The distal and proximal electrodes of the polyimide filament are connected to the electronic circuit through two dcblocking capacitors, which are aimed to prevent dc currents that can damage both the tissues and the electrodes [41]. A bridge rectifier based on four Schottky diodes (RB521ZS-30 by ROHM Co., Ltd.) provides full-wave rectification for the HF current picked up by the intramuscular electrodes. A limiting resistor followed by a Schottky diode (D1), and a set of three 2.2µF capacitors connected in parallel (Csleep) provide a smoothed input for a low-dropout linear regulator (ADP7112ACBZ-2.5 by Analog Devices, Inc.). The output of this regulator is connected to a set of two 10µF capacitors connected in parallel. This provides a stable dc voltage for the control unit and the rest of the electronics during and inbetween HF bursts. Sets of capacitors connected in parallel are used instead of single capacitors to ensure a proper capacitance and voltage rating in a miniature surface-mount package. Control unit: The wireless electronic circuit is controlled by an ultra-low power microcontroller (MKL03Z32CAF4R by NXP Semiconductors N.V.) with an Arm Cortex M0+ core. This 2mm × 1.6mm microcontroller is larger than that of [27], but includes several peripherals required by the circuit, a much lower power-up time, several low-power modes, and more general-purpose input/ output (GPIO) pins. The integrated high-speed comparator is used as the final stage of the downlink demodulator (explained below), and its output is connected to the integrated low-power UART receiver for further decoding. The UART transmitter generates the modulating signal for the load modulator used for uplink communications. Two digital outputs of the GPIO pins are used to control the switches of the current limiters for electrical stimulation, and the analog-to-digital converter (ADC) is used as the last stage of the AFE. Both circuits are described below. Finally, a low-power timer is used to control the sampling times during sensing, and to control the timeouts of the software. The control unit is configured to have three types of power consumption states: (1) idle mode, when the implant is waiting for commands from the external system, (2) processing and basic operations mode, when the implant decodes and processes the information coming from the external system and performs uplink communications and stimulation, and (3) sensing mode, when the implant is acquiring EMG activity and most of the Page 8 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 peripherals of the microcontroller are powered. The change of state is defined by the information received from the external system during downlink. The control unit is usually kept in a very-low power standby state (i.e., idle mode) to maintain the wireless circuit energized, avoiding the need to deliver a long Power up HF burst to power-up the microcontroller in-between bursts. Since the circuit lacks an active power source as a battery, it naturally shuts down at any time by disabling the bursts of HF current applied by the external system. The initialization byte sent by the external system is used by the control unit as the source for a hardware-based automatic wake-up matching, to wake up from the idle mode and go into the processing and basic operations mode. In the case the byte received matches the wake-up code already defined in the firmware, the wireless device wakes up and waits for the next byte with information to decode. The information includes the address of the wireless device or group of devices, and the command sent to it. Downlink demodulation and uplink modulation: For downlink, a demodulator was designed consisting of two voltage dividers whose outputs are connected to the inputs of a high-speed comparator integrated in the Fig. 5 Wireless device composed of intramuscular electrodes and miniature electronic circuit. a Top: circuit board (length: 10 mm; width: 8.5 mm), with four headers for their connection to the four electrode contacts of the intramuscular electrodes. Bottom: lateral view of the circuit (height: 2.4 mm). b Final conformation of the wireless device, including intramuscular electrodes and capsule protecting circuit. The guiding filament of the electrodes is inserted in the lumen of the insertion needle. c Circuit housed in a polymer capsule, connected to the intramuscular electrodes using a small PCB. d Image of rabbit’s hindlimb with two wireless devices implanted in the tibialis anterior and gastrocnemius medialis muscles. e Basic architecture of the wireless circuit shown in (a) Page 9 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 control unit. The output of the first voltage divider corresponds to the envelope of the downlink signal modulated in the HF current burst. The second voltage divider, which is connected immediately after the Schottky diode D1, behaves with the set of 2.2µF capacitors as a lowpass filter, creating a threshold proportional to the output of the first voltage divider. In other words, even if the amplitude obtained across the intramuscular electrodes varies due to fluctuations in the electric field applied by the external system, the outputs of both voltage dividers maintain their proportionality. The downlink signal is used for (1) receiving data at a rate of 256kbps or (2) receiving triggers to wake up the control unit using the synchronization byte or perform a function as uplink communications. The uplink is based on load modulation using on–off keying. The external system applies a HF current burst, which is detected by the wireless devices to do their specific uplink sequence. Additional file 1: Methods describes the uplink sequence as well as the timings required for each type of reply (e.g., ACK, send configuration, send sample). The floating device identifies this burst, starting the uplink section shown in Fig.3. Its control unit generates a 256kbps Manchester-encoded modulating signal with its UART transmitter. This modulating signal is used to command a set of transistor switches that short-circuit the inputs of the full-wave rectifier (Fig.5e). This results in virtually short-circuiting Electrodes ‘A’ and ‘B’, thus modulating the current consumption of the wireless device. These minute variations in consumption are seen in the sensing resistor of the external system during the uplink section (Fig.3, red waveforms), and are then demodulated and decoded by the digital unit. Load modulation is completely innocuous to the tissue: there is no electrical stimulation generated when uplink is performed. As a risk mitigation strategy, the load modulator is connected to the intramuscular electrodes through the dc-blocking capacitors. EMG analog front-end (AFE): One of the most challenging characteristics of the wireless device proposed here is the use of only two electrodes (Electrodes ‘A’ and ‘B’) for powering, bidirectional communications, electrical stimulation, and EMG sensing. As the wireless device powers from the volume-conducted alternating HF currents, very little consumption imbalances during the positive and negative semicycles may slightly charge the electrodes during these episodes, saturating the EMG amplifier. For this reason, the input of the EMG amplifier avoids the use of high dc-blocking capacitances that could prolong such saturation beyond the burst. Figure 6 shows the basic architecture of the designed EMG amplifier. A four-resistor difference amplifier with a gain of 18.4dB was designed based on a micropower operational amplifier (ADA4505 by Analog Devices, Inc). The amplifier is biased using a split resistor (RF/2), and a capacitor is added to each amplifier input, to act with the input resistors (RIN) as a first low-pass filter. A Zener diode clipper is included in both inputs of the operational amplifier for protection. The difference amplifier is followed by cascaded active band-pass filters with a bandwidth of 1kHz. The overall gain of the AFE is 54dB, and its output is connected to the ADC of the control unit, which is configured at a resolution of 10bits. During EMG sensing, the control unit is constantly monitoring when a HF burst is applied by the external system using the high-speed comparator. When this happens, depending on the sampling frequency set by the external system, the control unit replaces the samples corresponding to this saturation with a constant value (e.g., five samples for a sampling frequency of 1ksps). The number of samples to replace is implicitly set by the external system when configuring the sampling rate of the sensing mode. The replacement of the samples facilitates to identify the moments in which the AFE was saturated when the complete recording is uploaded to the external system, without misinterpreting the saturation with high amplitude EMG activity. Electrical stimulation: As mentioned above, the electrical stimulation mechanism of the wireless devices is based on the rectification of the volume conducted HF current bursts to cause local low frequency currents capable of stimulation. This is performed using two independent current limiters, each one connected to a Schottky diode (RB521ZS-30 by ROHM Co., Ltd.) that is connected to an electrode (Electrode ‘A’ or ‘B’) by means of a dc-blocking capacitor (Fig.5e). The architecture of the current limiter is explained in depth in Additional file1: Methods. Each current limiter is connected/disconnected from the load (i.e., the tissue) using a switch Fig. 6 Basic architecture of the AFE designed for the wireless device. The input of the AFE is the same pair of electrodes used for powering, communication and electrical stimulation Page 16 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 length: 3.8mm; surface area: 27 mm2). Despite this limitation, it was possible to accomplish a sufficient surface area to obtain enough electric current to power the wireless miniature electronic circuit, and to perform safe electrical stimulation without exceeding the charge injection capacity of the electrodes. Another substantial challenge was to combine the low-voltage EMG recordings with the high-voltage HF bursts for power. The WPT approach proposed here requires that the external system continuously delivers bursts of HF current for powering the wireless devices. The wireless device uses the same electrodes for powering, communications, EMG sensing and electrical stimulation. This implies that its AFE saturates when the HF bursts are applied by the external system. Despite this important limitation, and that the presence of the saturation of the AFE is seen in the raw recordings (flat lines in-between the EMG activity), the AFE designed for the circuit was able to have ample time windows (~ 14ms) to properly obtain EMG activity, ensuring a fast recovery from the saturation generated with the HF currents. These time windows will allow to record EMG signals with enough quality for their use in the BHNS proposed. Although the EMG recordings obtained were made with the hindlimb held to the load cell setup, it is intended that the system can be used in future acute human trials with free moving individuals, as the textile electrodes can be strapped firmly around the human limbs. The WPT approach uses a single channel (the tissues) both for powering and communication. This implies that while information is sent from a wireless device to the external system (uplink), other wireless devices cannot send information, nor the external system can do downlink. Another critical aspect of the approach proposed is that the semi-implantable wireless devices use only two electrodes for powering, bidirectional communications, EMG acquisition and electrical stimulation, limiting the possibility of performing several actions simultaneously. However, this characteristic favors the integration of the future ultrathin implantable devices (Fig.1). Additional file 1: Table S6, compares different implantable EMG sensors reported in the literature. Remarkably, all the works identified make use of WPT by inductive coupling. There are two typical conformations: (1) a central unit connected to the electrodes using leads (e.g., the IST-12 [7], the MyoPlant [15, 47, 48] and Ripple [49]), and (2) a cluster of wireless cylindrical implants (IMES [50]). This last conformation is very convenient, as the implantation procedure can be done using minimally invasive techniques [51]. Similarly, we envision our wireless devices as cylindrical and flexible implants that can be deployed by injection. These implants would have a hermetic housing protecting the ASIC with the circuit architecture proposed here. The architecture has successfully demonstrated the possibility to acquire EMG activity, with amplification factor, bandwidth and resolution similar to that of other implantable devices (Additional file1: TableS6). This can be useful for applications as neural interfaced assistive wearable robots for SCI, in which the control is done using intramuscular EMG-driven modelling [52]. The architecture has also demonstrated the possibility to perform electrical stimulation, opening the possibility to use the devices in multiple applications, including tremor management in essential tremor and Parkinson’s disease, and prosthetics control with sensory feedback [53]. Another WPT method that is gaining importance due to the possibility of obtaining small form factors is ultrasonic acoustic coupling. Very recently it was demonstrated the possibility of sending information from implantable “motes” to an external system for the use of the implants as neural recorders [54]. Yet the demonstrated implants do not include a hermetic capsule required for long-term implantation [18]. The capsule would increase the size of the mote and could attenuate the ultrasounds, imposing a critical constraint regarding the amount of energy obtained by the implant electronics. Ultrasonic WPT has also demonstrated higher penetration depth compared to inductive and capacitive coupling [24]. However, this is done by means of beam focusing, making it more difficult to arrange the external transmitter for powering and communicating with networks of wireless devices arranged through the body. Another drawback of ultrasonic acoustic coupling is the need to use gel for coupling the external transceiver and the skin [55]. This may cause wounds because of humidity, irritation due to allergens [56], and may be uncomfortable for chronic applications. Our WPT approach based on volume conduction avoids these drawbacks. In [57] it was determined that it would be possible to use existing rechargeable portable batteries (> 100Wh/kg) for the external system, accomplishing a portable unit that can be easily carried by patients. Also, the textile electrodes proposed do not require the use of gel and can be easily integrated in garments, making it more comfortable for the user, facilitating donning and doffing. Implant depth is a critical parameter for WPT methods. Here the wireless device was tested in vitro by placing the intramuscular electrodes at a depth of 3.25cm, and it could perform all the functions commanded from the external system. This ideal scenario did not include different conductivity layers as those Page 17 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 that could be present due to other tissues. However, we have in silico demonstrated using a multilayered geometry, that the electric field generated by the HF current bursts delivered by the external system are coarsely uniform in the region located between the external electrodes [30]. More importantly, we recently demonstrated in arms and lower legs of healthy humans that electric powers above 2mW and 5mW respectively could be obtained using needle electrodes (diameter: 0.4mm, length: 3mm) implanted approximately 1.75cm deep [31]. Therefore deeply implanted devices could be powered with this WPT approach. Conclusions The present work reports the development and successful evaluation of a technology composed of an external system that powers and controls wireless semi-implantable devices using a WPT approach based on volume conduction. Because the currents applied by the external system were applied in the form of bursts, they were below the heating limits defined by safety standards. The intramuscular electrodes proposed were appropriate for picking up the HF current bursts delivered by the external system to power and operate the designed miniature circuit. The semi-implantable devices for EMG sensing and stimulation could be configured and controlled from the external system using a bidirectional communications protocol that minimizes the application of HF currents. To the best of our knowledge, these are the first wireless devices powered by a WPT approach based on volume conduction that can do electrical stimulation and EMG sensing, and that bidirectionally communicate with the external system. It opens the path to the development of a BHNS that can do distributed electrical stimulation and sensing for neuroprostheses. Abbreviations BHNS: Bidirectional Hyper‑Connected Neural Systems; AIMDs: Active implant‑ able medical devices; EMG: Electromyography; SCI: Spinal cord injury; WPT: Wireless power transfer; HF: High frequency; ASICs: Application‑specific integrated circuits; UART : Universal asynchronous receiver‑transmitter; ASK: Amplitude‑shift keying; AFE: Analog front‑end; OSI: Open System Intercon‑ nection; SAR: Specific absorption rate; ACK: Acknowledge; PCB: Printed circuit board; FEM: Finite element method; SPICE: Simulation Program with Integrated Circuit Emphasis; GPIO: General‑purpose input/output; ADC: Analog‑to‑digital converter; LPF: Low‑pass filter; CMCiB: Centre for Compara‑ tive Medicine and Bioimage; IGTP: Germans Trias i Pujol Research Institute; TA: Tibialis anterior; GA: Gastrocnemius medialis. Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12984‑ 022‑ 01033‑3. Additional file1. Appendix: supplementary document with additional information regarding methods and results. This additional information is indicated on the body of the research article. Additional file2. Video with implantation procedure. Additional file3. Video showing the screenshot of the external system during bidirectional communications in an anesthetized rabbit. Acknowledgements The authors would like to express their gratitude to the team at Centre for Comparative Medicine and Bioimage (CMCiB) of the Gemans Trias i Pujol Research Institute (IGTP) for their work regarding the animal procedures. Authors’ contributions LBF, JM, AS and AI conceived and designed the study. LBF, JM, CR and AI developed the electronic hardware and the software. MOK, CW, AS and AI designed and developed the intramuscular electrodes. LBF and JM conducted the experiments, performed data analysis and drafted the manuscript. MOK, CR, CW, MTP, AC, FOB, AS and AI revised the manuscript critically. All authors read and approved the final manuscript. Funding This work has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 779982 (Pro‑ ject EXTEND—Bidirectional Hyper‑Connected Neural System), and from the European Research Council (ERC)—European Union’s Horizon 2020 research and innovation programme under Grant agreement No. 724244 (eAXON). CR has been also partially funded by CSIC Interdisciplinary Thematic Platform (PTI+) NEURO‑AGINGl+ (PTI‑NEURO‑AGING+). AI gratefully acknowledges the financial support by ICREA under the ICREA Academia programme. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The animal procedure was approved by the Ethical Committee for Animal Research of the Centre for Comparative Medicine and Bioimage (CMCiB) of the Germans Trias i Pujol Research Institute (IGTP), and by the Catalan Govern‑ ment (Project number: 10109). Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Author details 1 Department of Information and Communications Technologies, Universitat Pompeu Fabra, 08018 Barcelona, Spain. 2 Fraunhofer Institute for Biomedical Engineering IBMT, 66280 Sulzbach, Germany. 3 Research Centre for Information and Communications Technologies, University of Granada, 18014 Granada, Spain. 4 Department of Signal Theory, Telematics and Communications, University of Granada, 18014 Granada, Spain. 5 Neural Rehabilitation Group, Cajal Institute, Spanish National Research Council (CSIC), 28002 Madrid, Spain. 6 Electronics, Automation and Communications Department, ICAI School of Engineering, Comillas Pontifical University, 28015 Madrid, Spain. 7 Serra Húnter Fellow Programme, Universitat Pompeu Fabra, 08018 Barcelona, Spain. Received: 14 February 2022 Accepted: 19 May 2022 References 1. Kilgore KL, Anderson KD, Peckham PH. Neuroprosthesis for individuals with spinal cord injury. Neurol Res. 2020. https:// doi. org/ 10. 1080/ 01616 412. 2020. 17981 06. Page 18 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 2. Yildiz KA, Shin AY, Kaufman KR. Interfaces with the peripheral nervous system for the control of a neuroprosthetic limb: a review. J Neuroeng Rehabil. 2020;17(1):43. https:// doi. org/ 10. 1186/ s12984‑ 020‑ 00667‑5. 3. Basu I, Graupe D, Tuninetti D, Shukla P, Slavin KV, Metman LV, et al. Patho‑ logical tremor prediction using surface electromyogram and accelera‑ tion: potential use in ‘ON–OFF’demand driven deep brain stimulator design. J Neural Eng. 2013;10(3):36019. 4. Pascual‑Valdunciel A, Gonzalez‑Sanchez M, Muceli S, Adan‑Barrientos B, Escobar‑Segura V, Perez‑Sanchez JR, et al. Intramuscular stimulation of muscle afferents attains prolonged tremor reduction in essential tremor patients. IEEE Trans Biomed Eng. 2021;68:1768–76. 5. Hunt AJ, Odle BM, Lombardo LM, Audu ML, Triolo RJ. Reactive stepping with functional neuromuscular stimulation in response to forward‑ directed perturbations. J Neuroeng Rehabil. 2017;14(1):54. https:// doi. org/ 10. 1186/ s12984‑ 017‑ 0266‑6. 6. Sensinger JW, Dosen S. A review of sensory feedback in upper‑limb prostheses from the perspective of human motor control. Front Neurosci. 2020;14:345. https:// doi. org/ 10. 3389/ fnins. 2020. 00345. 7. Hart RL, Bhadra N, Montague FW, Kilgore KL, Peckham PH. Design and testing of an advanced implantable neuroprosthesis with myoelectric control. IEEE Trans Neural Syst Rehabil Eng. 2011;19(1):45–53. http:// ieeex plore. ieee. org/ docum ent/ 55858 27/. 8. Bergmeister KD, Hader M, Lewis S, Russold M‑F, Schiestl M, Manzano‑ Szalai K, et al. Prosthesis control with an implantable multichannel wire‑ less electromyography system for high‑level amputees. Plast Reconstr Surg. 2016;137(1):153–62. 9. Memberg WD, Polasek KH, Hart RL, Bryden AM, Kilgore KL, Nemu‑ naitis GA, et al. Implanted neuroprosthesis for restoring arm and hand function in people with high level tetraplegia. Arch Phys Med Rehabil. 2014;95:1201–11. 10. Kane MJ, Breen PP, Quondamatteo F, ÓLaighin G. BION microstimulators: a case study in the engineering of an electronic implantable medical device. Med Eng Phys. 2011;33(1):7–16. 11. Merrill DR, Lockhart J, Troyk PR, Weir RF, Hankin DL. Development of an implantable myoelectric sensor for advanced prosthesis control. Artif Organs. 2011;35(3):249–52. https:// doi. org/ 10. 1111/j. 1525‑ 1594. 2011. 01219.x. 12. Schulman JH. The feasible FES system: battery powered BION stimulator. Proc IEEE. 2008;96(7):1226–39. 13. Dinis H, Mendes PM. A comprehensive review of powering methods used in state‑of‑the‑art miniaturized implantable electronic devices. Biosens Bioelectron. 2021;172:112781. 14. Farnsworth BD, Triolo RJ, Young DJ. Wireless implantable EMG sensing microsystem. In: 2008 IEEE sensors. IEEE; 2008. p. 1245–8. http:// ieeex plore. ieee. org/ docum ent/ 47166 69/. Accessed 8 Sept 2017. 15. Ng KA, Rusly A, Gammad GGL, Le N, Liu S‑C, Leong K‑W, et al. A 3‑Mbps, 802.11g‑based EMG recording system with fully implantable 5‑elec‑ trode EMGxbrk acquisition device. IEEE Trans Biomed Circuits Syst. 2020;14(4):889–902. 16. Lee J, Leung V, Lee A‑H, Huang J, Asbeck P, Mercier PP, et al. Neural recording and stimulation using wireless networks of microimplants. Nat Electron. 2021;4(8):604–14. 17. Loeb GE, Peck RA, Moore WH, Hood K. BION™ system for distributed neural prosthetic interfaces. Med Eng Phys. 2001;23(1):9–18. 18. Piech DK, Johnson BC, Shen K, Ghanbari MM, Li KY, Neely RM, et al. A wireless millimetre‑scale implantable neural stimulator with ultrasonically powered bidirectional communication. Nat Biomed Eng. 2020;4(2):207– 22. https:// doi. org/ 10. 1038/ s41551‑ 020‑ 0518‑9. 19. Chang TC, Weber MJ, Charthad J, Baltsavias S, Arbabian A. End‑to‑ end design of efficient ultrasonic power links for scaling towards submillimeter implantable receivers. IEEE Trans Biomed Circuits Syst. 2018;12(5):1100–11. 20. Sonmezoglu S, Fineman JR, Maltepe E, Maharbiz MM. Monitoring deep‑ tissue oxygenation with a millimeter‑scale ultrasonic implant. Nat Bio‑ technol. 2021;39(7):855–64. https:// doi. org/ 10. 1038/ s41587‑ 021‑ 00866‑y. 21. Sedehi R, Budgett D, Jiang J, Ziyi X, Dai X, Hu AP, et al. A wireless power method for deeply implanted biomedical devices via capaci‑ tively coupled conductive power transfer. IEEE Trans Power Electron. 2021;36(2):1870–82. 22. Sodagar AM, Amiri P. Capacitive coupling for power and data telemetry to implantable biomedical microsystems. In: 2009 4th international IEEE/ EMBS conference on neural engineering. 2009. p. 411–4. 23. Hossain ANMS, Erfani R, Mohseni P, Lavasani HM. On the non‑idealities of a capacitive link for wireless power transfer to biomedical implants. IEEE Trans Biomed Circuits Syst. 2021;15(2):314–25. 24. Barbruni GL, Ros PM, Demarchi D, Carrara S, Ghezzi D. Miniaturised wire‑ less power transfer systems for neurostimulation: a review. IEEE Trans Biomed Circuits Syst. 2020;14(6):1160–78. 25. Turner BL, Senevirathne S, Kilgour K, McArt D, Biggs M, Menegatti S, et al. Ultrasound‑powered implants: a critical review of piezoelectric material selection and applications. Adv Healthc Mater. 2021;10(17):2100986. 26. Agarwal K, Jegadeesan R, Guo Y‑X, Thakor NV. Wireless power trans‑ fer strategies for implantable bioelectronics. IEEE Rev Biomed Eng. 2017;10:136–61. http:// ieeex plore. ieee. org/ docum ent/ 78798 07/. 27. Becerra‑Fajardo L, Schmidbauer M, Ivorra A. Demonstration of 2‑mm‑ thick microcontrolled injectable stimulators based on rectification of high frequency current bursts. IEEE Trans Neural Syst Rehabil Eng. 2017;25(8):1343–52. http:// ieeex plore. ieee. org/ docum ent/ 77260 54/. 28. Ivorra A, Becerra‑Fajardo L, Castellví Q. In vivo demonstration of injectable microstimulators based on charge‑balanced rectification of epidermically applied currents. J Neural Eng. 2015;12(6):66010. http:// stacks. iop. org/ 1741‑ 2552/ 12/i= 6/a= 066010. 29. Tudela‑Pi M, Becerra‑Fajardo L, García‑Moreno A, Minguillon J, Ivorra A. Power transfer by volume conduction: in vitro validated analytical models predict DC powers above 1 mW in injectable implants. IEEE Access. 2020;1. 30. Tudela‑Pi M, Minguillon J, Becerra‑Fajardo L, Ivorra A. Volume conduction for powering deeply implanted networks of wireless injectable medical devices: a numerical parametric analysis. IEEE Access. 2021;9:100594–605. 31. Minguillon J, Tudela‑Pi M, Becerra‑Fajardo L, Perera‑Bel E, Del‑Ama AJ, Gil‑Agudo Á, et al. Powering electronic implants by high frequency vol‑ ume conduction: in human validation. bioRxiv. 2021;2021.03.15.435404. https:// doi. org/ 10. 1101/ 2021. 03. 15. 43540 4v4. 32. Becerra‑Fajardo L, Ivorra A. First steps towards an implantable electro‑ myography (EMG) sensor powered and controlled by galvanic coupling. In: World congress on medical physics and biomedical engineering 2018 IFMBE proceedings. Springer Verlag; 2019. p. 19–22. 33. IEEE Standard for safety levels with respect to human exposure to elec‑ tric, magnetic, and electromagnetic fields, 0 Hz to 300 GHz. IEEE; 2019. 34. International Commission on Non‑Ionizing Radiation (ICNIRP). Guidelines for limiting exposure to Electromagnetic Fields (100 kHz to 300 GHz). Health Phys. 2020;118(5):483–524. 35. Muceli S, Poppendieck W, Hoffmann K‑P, Dosen S, Benito‑León J, Bar‑ roso FO, et al. A thin‑film multichannel electrode for muscle record‑ ing and stimulation in neuroprosthetics applications. J Neural Eng. 2019;16(2):026035. 36. Rose TL, Robblee LS. Electrical stimulation with Pt electrodes. VIII. Elec‑ trochemically safe charge injection limits with 0.2 ms pulses (neuronal application). IEEE Trans Biomed Eng. 1990;37(11):1118–20. 37. Poppendieck W, Sossalla A, Krob M‑O, Welsch C, Nguyen TAK, Gong W, et al. Development, manufacturing and application of double‑sided flex‑ ible implantable microelectrodes. Biomed Microdevices. 2014;16(6):837– 50. https:// doi. org/ 10. 1007/ s10544‑ 014‑ 9887‑8. 38. Muceli S, Poppendieck W, Negro F, Yoshida K, Hoffmann KP, Butler JE, et al. Accurate and representative decoding of the neural drive to muscles in humans with multi‑channel intramuscular thin‑film electrodes. J Physiol. 2015;593(17):3789–804. https:// doi. org/ 10. 1113/ JP270 902. 39. Poppendieck W, Dörge T, Hoffmann KP. Optimization of microporous plat‑ inum coatings for neural microelectrodes. In: 13th annual international conference of the IFES society. 2008. p. 319–21. 40. Meyer J‑U, Stieglitz T, Scholz O, Haberer W, Beutel H. High density inter‑ connects and flexible hybrid assemblies for active biomedical implants. IEEE Trans Adv Packag. 2001;24(3):366–74. http:// ieeex plore. ieee. org/ docum ent/ 938305/. 41. Cogan SF. Neural stimulation and recording electrodes. Annu Rev Biomed Eng. 2008;10(1):275–309. 42. Gabriel C, Gabriel S. Compilation of the dielectric properties of body tissues at RF and microwave frequencies. 1996. http:// niremf. ifac. cnr. it/ docs/ DIELE CTRIC/ Report. html. Page 19 of 19 Becerra‑Fajardoetal. Journal of NeuroEngineering and Rehabilitation (2022) 19:57 • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year • At BMC, research is always in progress. Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research ? Choose BMC and benefit from: ? Choose BMC and benefit from: 43. Silva A, Campos S, Monteiro J, Venâncio C, Costa B, Guedes de Pinho P, et al. Performance of anesthetic depth indexes in rabbits under propofol anesthesia: prediction probabilities and concentration‑effect relations. Anesthesiology. 2011;115(2):303–14. https:// doi. org/ 10. 1097/ ALN. 0b013 e3182 22ac02. 44. Peeters ME, Gil D, Teske E, Eyzenbach V, Brom WEVd, Lumeij JT, et al. Four methods for general anaesthesia in the rabbit: a comparative study. Lab Anim. 1988;22(4):355–60. 45. Spaich EG, Andersen OK, Arendt‑Nielsen L. Tibialis anterior and soleus withdrawal reflexes elicited by electrical stimulation of the sole of the foot during gait. Neuromodul Technol Neural Interface. 2004;7(2):126–32. https:// doi. org/ 10. 1111/j. 1094‑ 7159. 2004. 04016.x. 46. Hasgall P, Di Gennaro F, Baumgartner C, Neufeld E, Lloyd B, Gosselin M, et al. IT’IS Database for thermal and electromagnetic parameters of biological tissues. Zurich; 2018. https:// itis. swiss/ virtu al‑ popul ation/ tis‑ sue‑ prope rties/. 47. Morel P, Ferrea E, Taghizadeh‑Sarshouri B, Audí JMC, Ruff R, Hoffmann K‑P, et al. Long‑term decoding of movement force and direction with a wire‑ less myoelectric implant. J Neural Eng. 2016;13(1): 016002. 48. Kampianakis E, Sharma A, Arenas J, Reynold MS. A dual‑band wireless power transfer and backscatter communication approach for real‑time neural/EMG data acquisition. IEEE J Radio Freq Identif. 2017;1(1):100–7. 49. McDonnall D, Hiatt S, Smith C, Guillory KS. Implantable multichan‑ nel wireless electromyography for prosthesis control. In: 2012 annual international conference of the IEEE Engineering in medicine and biology society. IEEE; 2012. p. 1350–3. http:// ieeex plore. ieee. org/ docum ent/ 63461 88/. 50. Weir RF, Troyk PR, DeMichele GA, Kerns DA, Schorsch JF, Maas H. Implant‑ able myoelectric sensors (IMESs) for intramuscular electromyogram recording. IEEE Trans Biomed Eng. 2009;56(1):159–71. http:// ieeex plore. ieee. org/ docum ent/ 46336 66/. 51. Pasquina PF, Evangelista M, Carvalho AJ, Lockhart J, Griffin S, Nanos G, et al. First‑in‑man demonstration of a fully implanted myoelectric sensors system to control an advanced electromechanical prosthetic hand. J Neurosci Methods. 2015;244:85–93. 52. Jung MK, Muceli S, Rodrigues C, Megía‑García Á, Pascual‑Valdunciel A, Del‑Ama AJ, et al. Intramuscular EMG‑driven musculoskeletal modelling: towards implanted muscle interfacing in spinal cord injury patients. IEEE Trans Biomed Eng. 2022;69(1):63–74. 53. Salminger S, Sturma A, Hofer C, Evangelista M, Perrin M, Bergmeister KD, et al. Long‑term implant of intramuscular sensors and nerve transfers for wireless control of robotic arms in above‑elbow amputees. Sci Robot. 2019;4(32):eaaw306. 54. Ghanbari MM, Piech DK, Shen K, Alamouti SF, Yalcin C, Johnson BC, et al. A sub‑mm3 ultrasonic free‑floating implant for multi‑mote neural record‑ ing. IEEE J Solid‑State Circuits. 2019;54(11):3017–30. 55. Singer A, Robinson JT. Wireless power delivery techniques for miniature implantable bioelectronics. Adv Healthc Mater. 2021;10:2100664. 56. Chasset F, Soria A, Moguelet P, Mathian A, Auger Y, Francès C, et al. Con‑ tact dermatitis due to ultrasound gel: a case report and published work review. J Dermatol. 2016;43(3):318–20. 57. Becerra‑Fajardo L, Garcia‑Arnau R, Ivorra A. Injectable stimulators based on rectification of high frequency current bursts: power efficiency of 2 mm thick prototypes. In: Ibáñez J, González‑Vargas J, Azorín JM, Akay M, Pons JL, editors. Converging clinical and engineering research on neurorehabilitation II. Cham: Springer International Publishing; 2017. p. 667–71. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations.