A bioelectric router for adaptive isochronous neurostimulation enables multipolar bioelectric stimulation from a single source

Abstract

Multipolar intracranial electrical brain stimulation (iEBS) is a method that has potential to improve clinical applications of mono- and bipolar iEBS, including deep brain stimulation (DBS) and sensory prosthetics. Current tools for multipolar iEBS can have high entry costs, lack flexibility in managing stimulation parameters and electrodes, and can include unnecessary clinical features. To enable novel multipolar iEBS research, we developed the Bioelectric Router for Adaptive Isochronous Neuro Stimulation (BRAINS) board. The BRAINS board is a cost-effective, customizable, and scalable device designed to facilitate multipolar iEBS experiments using electrode arrays. The BRAINS board allows user configuration of each channel independently and prioritizes ease of integration with experimental setups. It supports remote configuration changes for rapid switching of electrode states while maintaining output isolation and low noise. We performed bench-top validation of monopolar, bipolar, and multipolar stimulation regimes as well as validation in vivo in mouse primary visual cortex and measured using Neuropixel recordings. The BRAINS board demonstrates no meaningful differences in Root Mean Square Error (RMSE) noise or signal-to-noise ratio compared to the baseline performance of the isolated stimulator alone. The board supports configuration changes at a rate of up to 600 Hz without introducing residual noise. The BRAINS board enables modulation of spatial and temporal specificity of electrical neuromodulation with stimulating arrays, with integration into control systems for real-time neural feedback.

Introduction

Electrical stimulation is one of a small number of clinically-tractable approaches to neuromodulation¹, along with focused ultrasound methods^{2, 3–4}. Electrical neuromodulation comes in a diversity of forms (transcranial direct current^5,6, transcranial alternating current⁷, intrancranial magnetic⁸, and intracranial⁹) and targets (peripheral nerves cochlea, spinal cord, and intracranial targets including the subthalamic nucleus, substantia nigra, motor cortex, sensory cortex¹⁰). Within the forms of electrical neuromodulation, intracranial electrical brain stimulation (iEBS) is a cornerstone therapeutic approach in clinical neurology and neurosurgery, particularly for treating Parkinson’s disease, with expanding indications for obsessive compulsive disorder^11,12, dystonia¹³, and epilepsy¹⁴. In addition to these expanding indications, clinical neural implants capable of iEBS are being tested as prospective sensory prosthetics^{15, 16–17} and for providing feedback to improve the quality and efficacy of motor brain-machine interfaces¹⁸. Furthermore, electrical microstimulation paradigms using low-amplitude and brief iEBS have been used for decades in animal research as a means of activating small numbers of neurons.

Despite widespread clinical use and its role in neuroscience research, our fundamental understanding of how iEBS affects neural circuits remains surprisingly limited. While we know that iEBS can activate varied neural elements—including somas, dendrites, and axons—the precise composition of recruited neural ensembles remains unclear^19,20. The general idea is that a small number of neurons in a uniform and symmetrical area around the stimulation site are activated; there is also ample evidence that the net effect of iEBS is suppressive, even if some neurons in the field are briefly activated. This knowledge gap is particularly evident in clinical settings, where stimulation parameters must often be modified empirically by clinicians during patient follow-up visits²¹, both to increase efficacy and to reduce side effects that arise through modulation of neural targets not intended to be modulated by the iEBS protocol. Multipolar brain stimulation, which extends iEBS to multiple spatially and temporally patterned electrical stimuli, has been proposed as a potential solution for increased iEBS control. Early efforts with small numbers of sites have shown promise²² in providing more precise^23,24 and efficient therapeutic outcomes^25,26.

Current FDA-approved devices typically offer unipolar, bipolar, or limited multipolar stimulation options up to 8 contacts²⁷. In practice, clinicians typically exhaust unipolar configurations before exploring more complex stimulation patterns¹⁹. Research devices with more contacts have been developed²⁸; whether more advanced iEBS approaches enabled by such density, such as multipolar pseudo-contacts²⁹ and current steering³⁰, which have been pioneered in peripheral neuromodulation, can add specificity³¹ to intracranial DBS-style devices is less clear. Investigation of the effects of advanced iEBS approaches,

[See PDF for image]

Fig. 1

Schematic of the BRAINS board. (A) General connection guide for full in vivo application of the board with general part guides and microcontroller requirements with the presence of an external output trigger from a stimulating device, the key components of the board that perform all the logic (computer output to microcontroller through octal latch multiplexer throw single pull 3 throw switches) and isolation (solid-state relay to split control signal from stimulator cathode/anode/ground signal connections to 16 output channels to connect to an implantable 16-channel electrode array), as well as a general setup of an in vivo invasive experiment (B) Modeled schematic of each electronic connection within the BRAINS board demonstrating control through a microcontroller to send signals to 4 octal latch multiplexers to 16 single pull 3 throw switches and where isolation occurs through optical isolation to activate and deactivate different channels with cathode, anode, or signal ground (C) 3D-rendered BRAINS board model with connectors (D) 3D-rendered BRAINS board model, including an integrated Raspberry Pi.

including complex multipolar stimulation, patterned stimulation, and other current steering paradigms faces significant technical and practical barriers in most research settings. The tools for exploring multipolar stimulation, such as clinical neurostimulators or high-end research systems, can be either prohibitively expensive, inflexible in their configuration, or excessive in their complexity for basic research applications²⁵. The research community currently lacks a flexible, open-source, cost-effective tool for exploring multipolar stimulation paradigms. These limitations have created a significant barrier to entry for researchers interested in exploring novel stimulation paradigms and understanding the basic principles of neural activation patterns.

To address these challenges, we developed a modular and readily reprogrammable interface—the Bioelectric Router for Adaptive Isochronous Neuro Stimulation (BRAINS) board for control of electrical brain stimulation. This system enables rapid switching and multipolar stimulation while maintaining compatibility with existing monopolar and bipolar stimulation experimental frameworks. Our approach prioritizes accessibility, flexibility, and signal isolation while allowing integration with standard experimental setups.

Methods

Device design and prototyping

We designed, fabricated, and evaluated a system for adaptive isochronous neurostimulation that enables software controlled selection channel state when using multisite electrical stimulation devices, such as electrode arrays. Functions of this device (Fig. 1A) include software selection of anodal and cathodal channels without needing to change physical connections, enabling multiple connections to anode and cathode channels, control of the state of non-used channels, and rapid switching between configurations. The BRAINS board was conceived and tested for use with silicon multi-channel electrodes (Fig. 1A, right, Neuronexus Technologies), but in principle can be coupled to any passive stimulation device.

The BRAINS board provides electronic switching capabilities for 16 independent electrode channels between four states (cathode, anode, signal ground, and floating) without requiring manual intervention during experiments while maintaining signal integrity. The board’s architecture (Fig. 1B) centers around four key interfaces: 1) A 2x20 female header compatible

Table 1. Control Electronics Properties. Timing and electrical characteristics of the Arduino, octal latch, and SP3T switch components indicating the theoretical minimum and maximum delays for rapidly switching channels due to every piece of hardware other than the solid-state relay.

[See PDF for image]

with both the Raspberry Pi 4B as a direct shield and adaptable to use with any Arduino (Fig 1C, D). Here, we present experiments and tests with the BRAINS board using an Arduino Pro Micro. 2) A complementary 2x20 male header providing access to auxiliary pins for external sensors, actuators, grounding, power, triggers, and access to the microcontroller’s built in +5 V, +3.3 V, and grounds (Fig. 1C,D) 3) Three standard banana connectors (Fig. 1C, D) to connect to any analog stimulus isolator: a red connector for positive terminal, black connector for negative terminal, and a green connector for connection to a signal or building ground. There is also a pad for a 1x1 header pin that is connected to the same ground. 4) Two standard 2x8 box connectors from Samtec (TSS-108-01-G-D) with dedicated building/signal ground connections to minimize noise, supporting flexible electrode configurations.

Signal routing and control are managed through multiple components, beginning with an octal transparent D-Type latch (Texas Instruments, CY74FCT373TSOC). This component provides stable channel selection through its latching capabilities while enabling efficient pin multiplexing. Operating at +5 V, each latch interfaces directly with the microcontroller’s logic levels with a theoretical maximum 8-nanosecond delay when latching per latch (Table 1). The multiplexing functionality reduces the number of required microcontroller connections to allow for scalability and customization options, such as daisy-chaining multiple boards or operating multiple BRAINS boards in parallel. The latch’s output and enable controls ensure precise timing of state changes across channels.

To ensure deterministic state selection, an analog SP3T triple-throw switch for each potential output (Texas Instruments, TS5A3357DCUR) routes signals between the four possible states (cathode, anode, signal ground, or floating). This switch has a theoretical maximum 7-nanosecond delay to support rapid state transitions for, in the case of the current design, 16 channels independently. (Table 1).

Table 2. Solid-State Relay Properties: Isolation, input/output, and delay characteristics of the solid-state relays on the BRAINS board as it theoretically affects our output signal through capacitance, leak, and limitations of the maximum voltage that can be sent from a analog isolated stimulating device.

[See PDF for image]

The final key component of the board design is the use of a solid-state relay (IXYS Systems, PAA193STR). These relays provide optical isolation between the control circuitry and stimulation pathways to prevent introduction of noise and maintain an isolated stimulation. The optical coupling mechanism prevents unwanted electrical interference while maintaining high-voltage handling capabilities at up to 600 V (Table 2) necessary for accommodating variable tissue and electrode impedance during current stimulation. The default configuration is “floating”, or open; this configuration minimized the potential for unintended stimulation through an electrode not selected as anode or cathode. The board includes a circuit for status indicators: a red LED power indicator and a green LED output enable indicator that illuminates when active channel state modifications are in progress.

We designed the BRAINS board using the Electronics Design function of Autodesk Fusion360; detailed descriptions of the design and files are available in our GitHub repository (denmanlab/BRAINSBoard, see Data and Materials availability section), including design files and 2-dimensional part footprints from each part’s documentation. The RoHS-compliant board was produced through a local third-party PCB manufacturing and assembly company (Colorado PCB Assembly) but can be produced by other PCB manufacturers using the provided files.

Software design and development

We designed two primary control interfaces for the BRAINS board: an Arduino-based serial communication protocol and direct Raspberry Pi GPIO control. These control interfaces do not both need to be used, but multiple control interfaces allow users with preferred approaches or variable comfort level with each approach to select the interface that works best for their existing neurophysiology system. The Arduino implementation utilizes a serial communication protocol operating at a baud rate of 115200 bits per second (bps), enabling precise temporal control of electrode states through a structured command syntax. Each electrode channel can be configured to one of the four states that the SP3T switches allow: floating, cathode, anode, or signal ground. State changes are managed through the octal latch system to ensure stable transitions.

The control architecture uses multiplexing to efficiently manage 16 channels through a minimal pin configuration. An output

[See PDF for image]

Fig. 2

Software Procedure for the BRAINS board. (A) Illustrated guide for input power for control devices to program individual channels with indications of what signal and latch enable outputs are required to enable a certain channel on an electrode. Each latch controls a set of 4 channels and the 8 signal pins are used in groups of 2 to determine which of four possible channel states is selected for each channel. Any channel is modified by setting the output enable HIGH, which allows each latch to modify its 8 output states, and configuring each latch enable signal output pins to reset the HIGH/LOW state for each signal input. LOW+LOW = floating, HIGH+LOW = cathode, LOW+HIGH = anode, and HIGH+HIGH= signal ground. (B) Basic programming guide using serial commands with indications of individual channel settings, varying forms of delays unique to electrical stimulation experiments, and how to setup loops within the framework of the Arduino Pro Micro code, with [] used to open and close setting the states, a hexadecimal value utilized to write to each individual channel, a concurrent letter corresponding to the channel state following, and key triggers between sets of [] to setup delays and loops for rapid switching purposes. 115200 bps baud rate through the Arduino serial command or Arduino serial command interface (e.g., PySerial). Full code available (see Data Availability Statement). (C) Arduino Pro Micro wiring connection schematic for setup direct to the BRAINS board. schematics for the Arduino setup of the BRAINS board in were designed using Fritzing CAD software. (D) Pinout for the control input 2x20 header (top) and signal output 2x8 headers (bottom) with bare minimum number of pins that are required to setup states for all individual channels as well as the exact order for channel number setup from BRAINS board to output pins. More detailed pinouts can be seen in Supplemental Fig. 1.

enable pin coordinates four latch enable pins to select channel groups (1–4, 5–8, 9–12, and 13–16). Eight data Input pins, connected across all octal latches, control the state transitions through a two-bit encoding scheme. This encoding determines the SP3T switch states, where the combinations of Input 1 and Input 2 (LOW-LOW: floating, HIGH-LOW: cathode, LOW-HIGH: anode, HIGH-HIGH: signal ground) define the channel configurations (Fig. 2A).

The serial command protocol implements a bracketed syntax e.g., ([...]) for channel configuration, supporting hexadecimal channel addressing e.g., (0-9, A-F) and state characters e.g., (F, C, A, G). Additional control features include external trigger synchronization (via pin A0), programmable delays (in seconds, milliseconds, or microseconds), and loop functionality for repeated patterns. This protocol structure enables complex stimulation sequences while maintaining precise timing control (Fig. 2B). The Arduino can easily be connected using simple male-to-male jumper wires (Fig. 2C) or can be connected to any alternative microcontroller using the drawn pinout (Fig. 4D).

As an alternative to this serial command protocol, a Raspberry Pi interface can provide direct GPIO control through a dedicated pin mapping, eliminating serial communication overhead. This implementation is particularly advantageous for applications that benefit from wireless control or require integration with complex input processing. The Raspberry Pi 4B’s GPIO pins 22–27 manage the BRAINS board output enable and latch enable functions, while pins 5, 6, 12, 13, 16, 17, 19, and 26 handle digital input control. This direct interface supports Python-based programming for flexible sequence generation and timing control through a precise sleep function.

Each control interface offers distinct advantages: the Arduino implementation is convenient for scenarios requiring minimal latency and integration with Windows-based instrumentation (e.g., Multi Channel Systems’ STG5 Isolated Analog Stimulator). The Raspberry Pi configuration is preferable for wireless control or complex input processing requirements. The choice between interfaces depends on experimental requirements regarding timing precision, communication flexibility, and system integration needs. Arduino or Raspberry Pi controls are intended to be used independently to suit experimenter preference, but can also be used in tandem if needed. The signal ground configuration connects any channel to a reference ground (accessed via the green banana connector), maintaining a consistent reference potential for electrophysiology and stimulation experiments. This comprehensive grounding scheme ensures the integrity of the signal in all operating modes. All code is open-source and public (see Data and Materials availability).

Bench-top validation

We performed two types of bench-top tests to match in vivo stimulation parameters: measurement of signal conditioning (Fig. 3A) and measurement of stimulation artifacts when stimulating through an electrode and recording independently through a neurophysiology recording electrode, in a conductive saline bath (Fig. 3B).

For signal conditioning measurements, we directly connected a high-power analog isolated stimulator (AM 4100, AM Systems) to the BRAINS board using a shielded pair of banana connectors and signal ground of the BRAINS board was grounded (to the building ground). Two 8-position, single-row connectors allowed us to connect output pins of the BRAINS board through a resistor (10 k, 100 k, 4x 100 k in series, or 1 M) to a recording systems (National Instruments PXI-6133 mutifunction IO card or Open Ephys Acquisition Board). We powered and controlled the BRAINS board with an Arduino Pro Micro; channel configurations were set using Arduino serial commands (Fig. 2B). An analog isolated stimulator sent custom waveforms (constant current or constant voltage) with variable shape (e.g., anodal monophasic, cathode and anode-leading biphasic), pulse durations, amplitude, and frequency for each pulse train. Stimulator output was coupled to the BRAINS board via a resistor for each individual channel (Fig. 3A). A parallel set of measurements, made by recording the outputs of the analog stimulus isolator through the same resistance but without routing through the BRAINS board, served as a control for the effect of BRAINS board routing.

For measurements in saline, we stimulated through a NeuroNexus A1x16 electrode array submerged in 1X Phosphate Buffered Saline (PBS) and recorded voltage with a Neuropixels 1.0 (see in vivo Validation for details on Neuropixels recording methods). We designed custom connectors using two 8-position single-row female connectors that were soldered directly to an Omnetics18 to free wire (A79045-001) connector, which directly connects to a NeuroNexus Adpt-A16-OM16 headstage. We performed impedance tests on all probes prior to experimentation using a dedicated impedance testing device (NanoZ, White Matter LLC) with an adapter to connect to the NeuroNexus array. During these tests we acquired continuous voltage series data through Neuropixels and Open Ephys GUI at 30 kHz and processed it using Open Ephys python tools and custom Python scripts.

In vivo validation

Ethical statement

All procedures using animals were approved by the University of Colorado Anschutz Institutional Animal Care and Use Committee (IACUC, protocol 00905) and were performed following these approved guidelines and regulations. The study was conducted in accordance with ARRIVE guidelines 2.0.

In vivo experimental procedures

C57BL6/J mice (n = 3 female, aged 3–11 months) initially underwent a surgical procedure to attach an aluminum head-fixation plate to the skull. In this part of the study, electrophysiological response of brain tissue to individual electrical pulses and pulse trains was compared all within one control experimental group, and therefore no randomization or blinding was employed. We selected the smallest sample size possible in order to validate the BRAINS board prototype in vivo for the first time in the present study. Mice were anesthetized with isoflurane (5%). The head-fixation plate was secured to the exposed skull using translucent Metabond dental cement. To seal the surgical site and facilitate later identification of lambda and bregma for sterotactic procedures, the translucent Metabond was applied to any remaining exposed skull. Following surgery, mice were given a 7-day recovery period before beginning head-fixation habituation. The habituation process involved gradually increasing the duration of head-fixation over 1–2 weeks until the mice exhibited no signs of distress during up to 2 h of head-fixation. Animals were included if habituation was successful; no animals were excluded.

Following habituation and directly before the electrophysiological recordings, mice were anesthetized and placed in a stereotaxic apparatus. Burr holes or small craniotomies were performed over the left visual cortex. Subsequently, mice were transitioned to a head-fixation platform on an in vivo electrophysiology rig and allowed to recover from anesthesia. Neuropixels 1.0 recording electrode(s) and a multichannel stimulating electrode (Neuronexus A1x16-5mm-50-703-A16, plated with IrOx for more effective current delivery) were inserted into the brain under piezoelectrical micromanipulator control (New Scale Technologies) at a rate of 50–100 m/min to a depth of greater than 1mm. The Neuropixels recording electrodes were inserted at a 45 angle to the brain surface and intersected the stimulating electrode, which was inserted at a 90 angle (vertically through the cortical depth), approximately 100-200m apart to prevent collisions. A stimulus isolator (AM4100, A-M Systems) was either (i) directly connected to the stimulating electrode via banana to mini-hookup clip attached to free wires from an Omnetics18 pin adapter (A79045-001, DigiKey) that mated with the A16 stimulating electrode adapter (Adpt-A16-OM16, Neuronexus) or (ii) routed through the BRAINS board via shielded banana to banana connectors. Charge-balanced bipolar and monopolar biphasic pulses with amplitudes varying from 100 to 100A were delivered through direct connection to stimulator and routed through the BRAINS board for direct comparison. Each parameter set was repeated 75 times with 2 secs between pulses. Following the completion of electrophysiological data collection, animals were euthanized following guidelines from the American Veterinary Medical Association.

Electrophysiological analyses

Neuropixels data was acquired using Open Ephys software; all data was processed and analyzed using custom Python code in Jupyter notebooks. Root mean square (RMS) calculations were averaged from 10 1-second chunks of the action potential (AP) band (hardware filtered, first order 300–6000 Hz) on a per-channel basis for both direct and BRAINS board connection setups. Prior to RMS measurement each voltage traces was de-spiked by removing values that exceeded 2.5 times the standard deviation of the mean voltage for each channel. For local field potential (LFP) signal processing, 10 seconds of LFP band (hardware filtered, first order 0.1–300 Hz) data were extracted, baseline-corrected to remove any direct current offset by subtracting the median voltage for each channel across the selected time points. Power spectral density (PSD) analysis between 0-100 Hz was conducted using the Welch method for each channel individually. The power values were then converted to a logarithmic scale (dB). The mean gamma-band power was calculated by averaging the PSD from 30–50 Hz for each channel and smoothed forplotting with a Gaussian filter ( = 2).

[See PDF for image]

Fig. 3

Stimulation Fidelity Bench Testing. (A) Schematic of direct current stimulation bench testing setup (B) Stimulation in saline through NeuroNexus A1x16 electrode and recording with a Neuropixels. Beaker image: Adobe Stock image. (C) Stimulation directly from isolator (black) vs through BRAINS board (red) with a 400 k resistor with varying stimuli at 100 s per phase cathode and anode leading square biphasic waveforms, demonstrating loss and shift in output waveform due to solid-state relay capacitance (D) Magnitude of peak-to-peak area under the curve for varying amplitudes across 400 ks for the BRAINS board and direct stimulation. For amplitudes of direct stimulation that were too large to measure (due to saturated amplifiers) a linear extrapolation of the direct measurement (red) is shown in dashed grey. Dashed lines highlight fixed gain reduction due to capacitance. (E) Peak-to-peak area under the curve charge balance for BRAINS board and direct stimulation parameters. Error bars are standard error of the mean.

Statistical tests were chosen based on the normality of the underlying data for each comparison throughout text—parametric tests (e.g., ANOVA) used when normal and non-parametric for non-normal data (Wilcoxon Rank Sum or Kruskal-Wallis). Error bars are standard error of the mean unless otherwise noted.

Results

Fidelity of stimulation

The fidelity of iEBS waveforms is critical to controlling the efficacy of iEBS. Any system needs to ensure high-fidelity waveforms, whether they are generated by in-built control circuitry or if the waveform passes through, as with the BRAINS board. To validate the fidelity of stimulation waveforms, we ran two types of validation tests: directly stimulating current through various resistors (Fig. 3A) and recorded by an analog signal acquisition system and stimulation through the stimulation electrode in saline and recording with a Neuropixel (Fig. 3B). There was a noticeable alteration in the signal output when routing with the BRAINS board (Fig. 3C), which we attribute to the output capacitance of the solid-state relays (50 pF, Table 2). We built an empirical comparison of the observed output magnitude to account for this capacitive drop by matching the peak to peak magnitude of current output over waveform time (Fig. 3D). Furthermore, the charge balance between the two devices is minimally different (p < 0.001, Cohen’s d < 0.2) within ranges that were recordable in direct resistor testing. (Fig. 3E). This is further explored with adjusted resistances of 100 k and 1 M resistors to demonstrate the differences between the direct current response compared to BRAINS board response in Supplemental Fig. 2.

Leak and channel isolation

We designed the BRAINS board to enable multipolar stimulation through independent channel control. Our evaluation focused on characterizing the temporal dynamics and isolation properties of the stimulation channels. We demonstrated the board switches both input and output configurations across all 16 channels at 689.7 Hz with no statistically significant noise caused by latency between control output and solid-state relay On/Off transitions (one sample t-test of the SNR of the Off Channels, p < 0.1) (Fig. 4A). To verify the contact switching speed, we delivered electrical pulses (n = 150) from an analog stimulus generator while alternating between two latch enable groups, each controlling 4 output signals. During active phases, we configured channels as either cathode or anode, defaulting to ground state when inactive, with latches 1 and 3 operating synchronously opposite to latches 2 and 4 (channel configuration details provided in Fig. 2A). Temporal characterization of the Arduino Pro Micro implementation showed that the output enable signal initialization for a 4-channel configuration through a single latch required approximately 146 s (mean = 46 s, sd=±1 sample for a 30 kHz recording rate), with each additional latch or 4-channel group requiring an incremental 67 s ±1 sample for state transitions during configuration switching (Fig. 4B). We did not observe significant cross-talk, with inactive (grounded) channels showing no detectable noise or cross-channel interference during rapid channel switching at 500 Hz stimulation frequencies (Fig. 4C). The recorded maximum and minimum voltages aligned precisely with theoretical predictions based on stimulation parameters (±5 V for 50 A biphasic stimulation across 100 k resistance), indicating robust channel isolation with no significant signal degradation beyond the capacitance effects previously addressed in Fig. 3D.

[See PDF for image]

Fig. 4

Limits and fidelity of rapid switching in bench testing. (A) Rapid testing (n = 75 trials per frequency) of leak through off channels and reduction of stimulus through on channels (100 s per phase cathode leading biphasic stimulation at 5V amplitude through 100 k resistance stimulation) while switching every channel from cathode-anode pair to ground. (B) Measured output enable delay due to software and Arduino Pro Micro delay,s for 1–4 latches enabled and switched. (C) Switching at 500 Hz over 16 output channels directly connected through 100 k resistors with 50 A cathode leading biphasic stimulation, with 500 s waveforms and inactive channels grounded. (D) Gaussian smoothed average plot of 300 channels of Neuropixels recording in saline, average voltage measurement over n = 75 trials for 50 A, cathode-leading biphasic stimulation through a cathode and multiple (colors) adjacent anodes. The threshold (mean + 2.5 s.d of channels 0–50 and 250–300) shown as dashed line; the contiguous number of channels above this threshold for each number of anodes shown inset. (E) Average channel distance of voltage spread with varying amplitudes of cathode leading biphasic current stimulation. (F) Average area under the curve of a 300 channel range Neuropixels recording of stimulation with varying amplitudes of cathode leading biphasic current stimulation. Error bars are standard error of the mean.

We evaluated the BRAINS board’s capacity to enable multipolar stimulation through an electrode array with a single analog stimulus isolated using a systematic characterization of observed stimulation field properties in saline. When implementing focused multipolar configurations with a single cathode and distributed anodes positioned 50 m above and below the stimulating electrode, we observed systematic enhancement of peak voltages with increasing anode count (Fig. 4D). Correlation analysis showed a strong positive relationship between anode count and maximum voltage (Spearman’s = 0.90, p < 0.05), with peak voltages increasing from 86.85 ± 3.21 V (2 anodes) to 112.74 ± 3.06 V (4 anodes) to 164.13 ± 3.12 V (6 anodes) 157.88 ± 2.97 V (8 anodes) 188.17 ± 3.07 V (10 anodes). The spatial extent of stimulation, quantified as the full width of the voltage profile above threshold, showed a corresponding positive trend with anode count ( = 0.87, p = 0.054), expanding from 76 to 109 channels. While individual configuration comparisons did not reach statistical significance for either maximum voltage or spread width (Kruskal-Wallis test, p = 0.41 for both metrics), the monotonic relationship suggested systematic modulation of both field strength and spatial distribution through anode count manipulation.

Analysis of stimulation spread across Neuropixels recording channels (2 channels per 10 m) revealed consistent spatial distributions across configurations (Fig. 4E). Statistical analysis demonstrated significant effects of both stimulation amplitude (Kruskal-Wallis H = 17.03, p < 0.001) and anode count (Friedman = 13.60, p < 0.01) on spread characteristics. Post-hoc analyses revealed that 100 A stimulation produced significantly broader spreads compared to 25 A (p < 0.01) and 5 A (p < 0.001) conditions, but not 50 A (p = 0.54). Both 50 A and 25 A conditions generated significantly larger spreads than 5 A stimulation (p < 0.001). While positive correlations between anode count and spread distance existed across all amplitudes ( = 0.70–0.80), these relationships did not achieve statistical significance (all p > 0.10). Area Under Curve (AUC) analysis across 300 channels suggested amplitude-dependent effects on total voltage distribution (Fig. 4F). This effect appeared most pronounced at 100 A, where AUC values ranged from 11.1 to 16.5 mVchannels and exhibited saturation with increasing anode count. The relationship between AUC and anode count followed similar patterns across all tested amplitudes (5 A, 25 A, 50 A, and 100 A), characterized by steep increases between 2 and 4 anodes followed by more gradual increases or plateaus. An ANOVA revealed significant amplitude effects (F = 63.14, p < 0.001), with post-hoc comparisons confirming hierarchical differences between all amplitudes except 25 A and 5 A (p = 0.078). Linear relationships between AUC and anode count achieved significance for 50 A and 25 A conditions (p < 0.05), while the 100 A condition demonstrated apparent saturation, likely due to recording system limitations.

In vivo testing

[See PDF for image]

Fig. 5

In vivo comparison of signal and noise with direct connection and BRAINS board. (a) schematic for in vivo electrophysiology with electrical stimulation setup (b) raw (left column, gray and pink traces) and despiked (right column, red and black traces) voltage traces from a single channel for direct connection (top, red) and BRAINS board (bottom, black). (c) paired despiked RMS with mean for direction connection and BRAINS board (n = 300 channels, Wilcoxon Test, p = 0.02). (d–f) raw LFP heatmap (d), LFP spectral power (e), and gamma power (f) for direct connection (top) and BRAINS board (bottom).

To validate the utility of the BRAINS board in experimental conditions, we inserted a linear 16-channel stimulating electrode and high-density electrophysiological recording array (Neuropixel) in mouse visual cortex (Fig. 5A). First, to ensure that connecting the stimulating electrode through the BRAINS board does not meaningfully increase the noise in the recording compared to direct connection, we measured the RMS of despiked high-pass voltage traces across channels (Fig. 5B, C). BRAINS board increased the mean RMS by 0.46 V (direct: 10.49 V ± 3.45, BRAINS board: 10.95 V ± 3.14, Wilcoxon test p < 0.01). This difference, while statistically significant, represents a negligible physiological difference and supports that the BRAINS board is not adding a meaningful source of electrical noise. Next, we qualitatively compared the LFP signal recorded during direct connection and connection through BRAINS board. The raw LFP signal (Fig. 5D), the power spectra of LFP (Fig. 5E), and the LFP power in the gamma band (30–50 Hz, Fig. 5F) across channels were all similar, demonstrating that connection to the stimulating electrode array through the BRAINS board does not alter local neurophysiology in the implanted tissue.

Next, we compared the stimulation efficacy through the BRAINS board compared to direct connection in vivo. We stimulated in mouse visual cortex while simultaneously recording the nearby artifact and evoked potential from a Neuropixel recording electrode. During prior benchtop testing, we established a near equivalent stimulation dose by using the waveform AUC measured at 400 k for direct connection and BRAINS board (Fig. 3D). The circuit resistance changes the dose relationship (Supplemental Fig. 2), and we selected 400 k resistance because it approximated the mean contact impedance of the stimulating electrode (407±31 k), and thus an approximation of the in vivo circuit. Using this relationship, we identified that the measured AUC of 100 A routed with the BRAINS board (24.054 ± 2.96) is less than 5 percent (4.8 percent) different from 25 A direct connection (25.01 ± 2.96) in the benchtop configuration. Thus we compared 100 A routed with the BRAINS board and 25 A direct connection, expecting less than a 5 percent difference. The electrophysiologically measured artifact and evoked potentials at a single channel (Fig. 6A, top) and the corresponding voltage heatmap for all inserted recording channels (Fig. 6B, bottom) are visually similar for the approximated equivalent currents. To quantitatively assess similarity, we compared the AUC for the recorded extracellular voltage (including artifact and subsequent evoked potentials) for both conditions. Routing stimulation through the BRAINS board statistically increased the AUC of recorded response (direct 25 A : 21.65 ± 2.3, BRAINS board 100 A : 22.48 ± 1.5, unequal variances t-test: p = 0.02), but remained less than 5 percent different (4.8 percent). Thus, while we note small differences, the BRAINS board can deliver current dosages within 5 percent of those delivered through direct connection, when accounting for electrode impedances. The single channel raw voltage traces and full probe voltage heatmaps for amplitudes 100, 50, 25, 5, 5, 25, 50, and 100 for bipolar and monopolar can be viewed in Supplemental Fig. 3, though attenuation prevented from aligning the smallest ampltiude (5 A) stimuli through the BRAINS board for proper comparison.

Finally, we assessed whether switching near the limit rates of the BRAINS board cause any residual stimulation artifacts when as well as whether switching the simulation site was effective in shifting the locus of stimulation. To do so, we implanted a 16-channel stimulating electrode such that all stimulating electrodes were within cortex (primary visual cortex), while measuring across the full depth of cortex. We then used the BRAINS board to switch the stimulating electrode from the deepest to the most superficial sequentially (Fig. 7a) at both 500 Hz (2 msec between each stimulation) and 40 Hz (25 msec between each stimulation). We compared the stimulation artifact during the most rapid switching epoch (500Hz) to amplitude and polarity matched single pulse stimulation, i.e. the same stimulus not in a rapid switching train. We did not observe a difference in stimulus artifact on the channel with the largest response during rapid switching (Fig. 7b, colored traces) compared to the mean single pulse (Fig 7b, red trace). This was not only true of the largest channel, but also across all channels in the cortex (Fig. 7c). During this 500 Hz switching scenario we observed a systematic shift in the field potentials in the response in the small windows between stimuli (Fig. 7c). To better observe spatial shifts in the evoked response we slowed the switch rate to 40 Hz to allow 25 msec windows, and observed clear shifts in the response potentials (Fig. 7d), demonstrating the potential for spatiotemporal multipolar stimulation to increase the spatial and temporal specificity of electrical neuromodulation.

[See PDF for image]

Fig. 6

In vivo Comparison of BRAINS Board and Direct Connection Stimulation. (a) single channel (channel = 150) raw voltage trace (top) and mean voltage heatmaps across all inserted channels (n = 75 trials) for matched current doses (25 A direct and 100 A BRAINS board) measured in vivo in mouse visual cortex. (b) AUC for measured in vivo extracellular voltage during electrical pulses for 25 A direct and 100 A BRAINS board (n = 75 trials for each condition, unequal variances t-test: p = 0.02). All error bars are standard error of the mean.

Discussion

The goal of the device reported here, the BRAINS board, is to allow simultaneous multipolar and/or rapidly switching electrical stimulation through high contact count devices, particularly in research settings. We demonstrate that the BRAINS board faithfully transmits arbitrary waveforms to any available electrode (Fig. 2), does not introduce electrical noise to either stimulation waveforms (Fig. 4C) or into the nearby extracellular space, and enables both multipolar stimulation and rapid switching. The BRAINS board interfaces with any lead or passive electrode array, and we validate its use in intracortical electrical stimulation in mice. Although the scope of validation for this prototype study (n=3) demonstrates effectiveness and allowed us to measure performance, a larger sample size will be required to demonstrate generalizability and more potential failure modes. While spatiotemporal multipolar stimulation through an electrode array could be achieved via independent control of each channel with dedicated stimulus isolator per-channel, by multiplexing a single stimulus isolator BRAINS board allows scalability to higher channel counts by either combining multiple BRAINS board or making our PCB larger. Further, we provide specifications, open-sourced designs, and software to enable manufacture of this device at or near materials cost.

Intracranial electrical brain stimulation, in both research settings (e.g., microstimulation^32,33) and clinical applications (e.g., deep brain stimulation⁹), typically relies on a stimulus isolator³⁴. Such isolators can be analog- or digitally-controlled, and rely on an optically-isolated battery powered circuit to generate constant voltage or constant current pulse to a single set of anode and cathode outputs. The BRAINS board does not replace this isolator in a neuromodulation system, but instead is positioned between an isolator and the stimulus effector in tissue (i.e., the lead or electrode) to allow flexibility in the routing of the limited anode and cathode outputs of stimulus isolators. One experiment the BRAINS board enables is the iterative exploration of the effect of single-site or all paired bipolar configuration allowed by any electrode geometry. This functionality is particularly useful for the rapidly expanding field of high density^{35, 36, 37–38}, usually silicon-based³⁹ neural microelectrodes. Because the BRAINS board is digitally-controlled (Fig. 2), when paired with a digitally-controlled stimulus isolator the spatiotemporal pattern of neuromodulation is limited only by the BRAINS board rate of switching (Fig. 4A-B). The BRAINS board facilitates novel application of neuromodulation through high-density electrode arrays, both extant and arising electrode technologies⁴⁰. The BRAINS board could also be applied to empirically validate complex stimulation protocols for peripheral neuromodulation devices⁴¹, where for some devices such as cochlear implants complex neuromodulation is known to have benefits^42,43, while in others such approaches remains not yet empirically vetted (e.g., vagus nerve stimulation⁴⁴).

By enabling the routing of stimulator outputs to any connected electrode, the BRAINS board allows nearly arbitrary spatiotemporal control of neuromodulation (e.g., multipolar or temporal interference stimulation protocols) with any lead or passive electrodes. The rapid switching of the BRAINs board moves towards the possibility of temporal interference with single source, however typical temporal interference protocols typically use greater kHz-range carrier frequencies^45,46. While we think it worthwhile to test the effectiveness of temporal interference near the maximum switching rate of the BRAINS board (500 Hz), this is at the lowest effective frequencies of temporal interferences⁴⁶. Future iterations of the BRAINS board will need to improve on the effective switching speed in order to enable higher carrier frequencies for temporal interference protocols. While the BRAINS board facilitates using such complex protocols with arbitrary electrodes, these protocols are not unique to the BRAINS board. One example is “current steering” through multiple contacts simultaneously³¹. The potential benefits of such protocols have been an area of active research for years. Notions of current steering for neuromodulation originated with theoretical and computational models^{47, 48–49}, with the control of stimulated volume (and therefore potential limitation of off-target effects) a primary proposed benefit of current steering. However, biological validation of these proposed benefits of current-steering has been difficult. The most robust testing has come in direct clinical studies^50,51, pre-clinical validation in animal neural tissue are limited⁵². Some pre-clincal behavioral detection studies show differences with current steering⁵³, but direct measurements of activated volume, and therefore insight into mechanism of action, are much more sparse⁵⁴. The BRAINS board will allow such measurement in neurophysiology labs. to push the potential, need to try other geometries, and the BRAINS board will allow current steering with higher density and other bespoke stimulating electrodes.

[See PDF for image]

Fig. 7

In vivo comparison of artifact shape and evoked potentials during rapid spatiotemporal switching with the BRAINS board. (a) schematic for in vivo pattern of sequenctial switching of stimulation across the stimulation array (b) Evoked artifacts at the peak channel during BRAINS board channel switching, both static non-switching stimulation at each channel (red) and at 500 Hz, (colors corresponding to switching to each channel, as in panel a). (c) evoked local field potential heatmap during 500 Hz switching up the stimulation electrode. Location of stimulating site shown in blue-green colors corresponding to panel a. (d–f) evoked local field potential heatmap during 40 Hz switching up the stimulation electrode. Location of stimulating site shown in blue-green colors corresponding to panel a.

In addition to current steering and directional stimulation through spatial patterning, temporal patterning is another frontier in neuromodulation⁵⁵. Temporal patterning of single stimulation sites, where these patterns are determined a priori, has strong impacts on deep brain stimulation (DBS) effectiveness, with proscribed patterns able to both increase⁵⁶ and eliminate⁵⁷ the therapeutic effectiveness. Biomimetic stimulation, where the temporal patterning is designed to mimic known statistics of the neural activity in the area being modulated, can also profoundly affect perception of neuromodulation for sensory prostheses⁵⁸. Both a priori temporal patterning and biomimetic patterning could be extended to include spatiotemopral “flow” of patterns across neural circuits with multiple electrodes. Research into the effectiveness and mechanisms of these protocols is enabled by the BRAINS board. Finally, “real-time” or “closed-loop”⁵⁹ control of neuromodulation based on neural or other feedback⁶⁰ to achieve an optimal modulation is a rapidly expanding form of neuromodulation. The combination of software control of rapid switching through the BRAINS board with complex neural readout available in research settings⁶¹, can shed light on relevant biomarkers for such closed-loop neuromoduation.

The BRAINS board’s capabilities enable novel studies of electrophysiological effects of iEBS stimulation. Clinical trials have highlighted that optimal stimulation parameters may not be intuitive, and may require computational identification⁶². While clinical systems deliver directional stimulation, they cannot readily facilitate systematic investigation of neural responses, especially with the single neuron resolution across populations and neural circuits needed to optimize these therapies. Clinical research shows that LFPs features such as beta oscillations could serve as biomarkers for stimulation optimization⁶³; studying how complex spatiotemporal stimulation patterns influence these population-level signals requires experimental flexibility. The BRAINS board fills this research gap by enabling precise control over stimulation parameters while allowing simultaneous electrophysiological recordings, making it possible to systematically map relationships between stimulation patterns and population dynamics. Such a systematic map, enabled by the BRAINS board, will enhance approaches to stimulation programming and could help resolve ongoing questions about how directionality, current steering, and temporal patterning influence therapeutic outcomes at the circuit level.

The BRAINS board, in its current form, enables research experiments into complex spatiotemporal neuromodulation and integration with neurophysiology tools to understand the mechanisms of neuromodulation. To continue to improve the capacities of the BRAINS board future developments the BRAINS board can be extended in future versions. Pass-through signal fidelity will be improved by upgrading the solid-state relays (e.g, to IXYS Systems, OAA160STR). This modification will reduce output capacitance from 50 pF to 5 pF on isolated channels and decrease leak current from 10 A to 250 nA, enhancing signal fidelity to 97% when working with high impedance electrodes. Ground loop interference remains a persistent challenge in electrophysiology experiments, particularly manifesting in the LFP band during serial command transmission. To address this, we propose integrating an embedded microprocessor directly onto the BRAINS board to establish a single ground reference point and implement comprehensive electrical isolation from computer interfaces. Alternatively, removal of Arduino dependency should notably reduce serial command-related noise in the LFP band while maintaining signal integrity across all connected components. Future development will focus on implementing modular architecture, with a base control module serving as the central processing unit and ground reference point, supplemented by attachable 16-channel shields for scalable expansion. This modular approach will facilitate integration with various neuroscience tools, including recording devices and optogenetic instruments. Several advanced features are under consideration for future iterations, including FPGA integration for closed-loop stimulation capabilities and a channel-specific resistance control system, enabled by per-channel impedance measurement circuit. This impedance measurement circuit will be achieved in the next generation through a single additional multiplexer to route of a fixed, board-generated input signal to each channel instead of the external inputs. These additions would enable more precise current delivery through individual electrodes while maintaining the efficiency of a single analog isolated stimulator. The proposed modular architecture ensures that these advanced features can be implemented without compromising the system’s core functionality or compatibility with existing experimental setups.

In conclusion, the BRAINS board enables novel software-based and near real-time control of electrical stimulation through any stimulating electrode, facilitating the preclinical study of complex spatiotemporal patterning of intra- and transcranial neuromodulation through electrode arrays. By enabling research into such patterning with emerging research devices, the BRAINS board will advance understanding of the basic mechanisms of neuromodulation and facilitate improvements in current approaches as well as novel technologies.

Acknowledgements

We would like to thank Grant Hughes for assistance with electrical stimulation methods used in this work and Moriah Miles for assistance with in vivo testing.

Author contributions

Conceptualization: ES, JH, DJD Methodology: ES, JH, DJD Investigation: ES, JH, DJD Visualization: ES, JH Supervision: DJD Writing-original draft: ES, JH, DJD Writing-review and editing: ES, JH, DJD.

Funding

This work was supported by: National Institutes of Health grant R01NS120850 (DJD).

Data availability

Data, software, and analysis code are available from Github, denmanlab/BRAINS board.

Declarations

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-025-07568-4.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Davidson, B et al. Neuromodulation techniques: From non-invasive brain stimulation to deep brain stimulation. Neurotherapeutics; 2024; 21, [DOI: https://dx.doi.org/10.1016/j.neurot.2024.e00330] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38340524][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11103220]e00330.https://www.sciencedirect.com/science/article/pii/S1878747924000163

2. Darmani, G et al. Non-invasive transcranial ultrasound stimulation for neuromodulation. Clin. Neurophysiol.; 2022; 135, pp. 51-73.1:STN:280:DC%2BB2M%2FosF2qtQ%3D%3D [DOI: https://dx.doi.org/10.1016/j.clinph.2021.12.010] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35033772]https://www.sciencedirect.com/science/article/pii/S1388245721008920

3. Legon, W. et al. A retrospective qualitative report of symptoms and safety from transcranial focused ultrasound for neuromodulation in humans. en. Scientific reports 10. Publisher: Nature Publishing Group, 5573. ISSN: 2045-2322. https://www.nature.com/articles/s41598-020-62265-8 (2020).

4. Dell’Italia, J., Sanguinetti, J. L., Monti, M. M., Bystritsky, A. & Reggente, N. Current state of potential mechanisms supporting low intensity focused ultrasound for neuromodulation. English. frontiers in human neuroscience 16. Publisher: Frontiers. ISSN: 1662-5161. https://doi.org/10.3389/fnhum.2022.872639/full (2022).

5. Woods, AJ et al. A technical guide to tDCS, and related non-invasive brain stimulation tools. Clin. Neurophysiol.; 2016; 127, pp. 1031-1048.1:STN:280:DC%2BC28vptFWktA%3D%3D [DOI: https://dx.doi.org/10.1016/j.clinph.2015.11.012] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26652115]https://www.sciencedirect.com/science/article/pii/S1388245715010883

6. Chase, HW; Boudewyn, MA; Carter, CS; Phillips, ML. Transcranial direct current stimulation: A roadmap for research, from mechanism of action to clinical implementation. Mol Psychiatry; 2020; 25, pp. 397-407. [DOI: https://dx.doi.org/10.1038/s41380-019-0499-9] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31455860]

7. Del Felice, A et al. Personalized transcranial alternating current stimulation (tACS) and physical therapy to treat motor and cognitive symptoms in Parkinson’s disease: A randomized cross-over trial. NeuroImage: Clin; 2019; 22, [DOI: https://dx.doi.org/10.1016/j.nicl.2019.101768] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30921609]101768.https://www.sciencedirect.com/science/article/pii/S2213158219301184

8. Barker, A. T., Jalinous, R. & Freeston, I. L. Non-invasive magnetic stimulaiton of human motor cortex. The Lancet. Originally published as Volume 1, Issue 8437 325, 1106–1107. ISSN: 0140-6736. https://www.sciencedirect.com/science/article/pii/S0140673685924134 (1985).

9. Lozano, A. M. et al. Deep brain stimulation: Current challenges and future directions. en. Nature Reviews Neurology 15. Publisher: Nature Publishing Group, 148–160. ISSN: 1759-4766. https://www.nature.com/articles/s41582-018-0128-2 (2019).

10. Krishna, V; Fasano, A. Neuromodulation: Update on current practice and future developments. Neurotherapeutics; 2024; 21, [DOI: https://dx.doi.org/10.1016/j.neurot.2024.e00371] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38734464][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11103215]e00371.

11. Greenberg, B. D. et al. Three-year outcomes in deep brain stimulation for highly resistant obsessive–compulsive disorder. en. Neuropsychopharmacology 31. Publisher: Nature Publishing Group, 2384–2393. ISSN: 1740-634X. https://www.nature.com/articles/1301165(2006).

12. Wichmann, T; DeLong, MR. Deep brain stimulation for neurologic and neuropsychiatric disorders. Neuron; 2006; 52, pp. 197-204.1:CAS:528:DC%2BD28XhtFentr%2FN [DOI: https://dx.doi.org/10.1016/j.neuron.2006.09.022] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17015236]https://www.sciencedirect.com/science/article/pii/S089662730600729X

13. Coubes, P et al. Electrical stimulation of the globus pallidus internus in patients with primary generalized dystonia: Long-term results. J. Neurosurg.; 2004; 101, pp. 189-194. [DOI: https://dx.doi.org/10.3171/jns.2004.101.2.0189] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15309907]

14. Heck, CN et al. Two-year seizure reduction in adults with medically intractable partial onset epilepsy treated with responsive neurostimulation: Final results of the RNS system pivotal trial. Epilepsia; 2014; 55, pp. 432-441. [DOI: https://dx.doi.org/10.1111/epi.12534] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24621228][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4233950]

15. Orlemann, C et al. Flexible polymer electrodes for stable prosthetic visual perception in mice. Adv. Healthcare Mater.; 2024; 13, 2304169.1:CAS:528:DC%2BB2cXltVGgurs%3D [DOI: https://dx.doi.org/10.1002/adhm.202304169]

16. Fernández, E. et al. Visual percepts evoked with an intracortical 96-channel microelectrode array inserted in human occipital cortex. en. The Journal of Clinical Investigation 131. Publisher: American Society for Clinical Investigation. ISSN: 0021-9738. https://www.jci.org/articles/view/151331 (2021).

17. Towle, V. L., Pham, T., McCaffrey, M., Allen, D. & Troyk, P. R. Toward the development of a color visual prosthesis. en. Journal of Neural Engineering 18. Publisher: IOP Publishing, 023001. ISSN: 1741-2552. https://doi.org/10.1088/1741-2552/abd520 (2021).

18. Ajiboye, AB et al. Restoration of reaching and grasping movements through brain-controlled muscle stimulation in a person with tetraplegia: A proof-of-concept demonstration. Lancet; 2017; 389, pp. 1821-1830. [DOI: https://dx.doi.org/10.1016/S0140-6736(17)30601-3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28363483][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5516547]

19. Vorwerk, J; Brock, AA; Anderson, DN; Rolston, JD; Butson, CR. A retrospective evaluation of automated optimization of deep brain stimulation parameters. J. Neural Eng.; 2019; 16, 2019JNEng.16f4002V [DOI: https://dx.doi.org/10.1088/1741-2552/ab35b1] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31344689][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7759010]064002.

20. Hartmann, CJ; Fliegen, S; Groiss, SJ; Wojtecki, L; Schnitzler, A. An update on best practice of deep brain stimulation in Parkinson’s disease. Ther. Adv. Neurol. Disord.; 2019; 12, 1756286419838096. [DOI: https://dx.doi.org/10.1177/1756286419838096] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30944587][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6440024]

21. Deuschl, G. et al. Deep brain stimulation: postoperative issues. eng. Movement Disorders: Official Journal of the Movement Disorder Society 21 Suppl 14, S219–237. ISSN: 0885-3185 (2006).

22. Buhlmann, J; Hofmann, L; Tass, PA; Hauptmann, C. Modeling of a segmented electrode for desynchronizing deep brain stimulation. Front Neuroeng; 2011; 4, 15.1:STN:280:DC%2BC38%2Fntlajtg%3D%3D [DOI: https://dx.doi.org/10.3389/fneng.2011.00015] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22163220][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3233722]

23. Anderson, DN et al. The DBS: Multiresolution, directional deep brain stimulation for improved targeting of small diameter fibers. Front. Neurosci.; 2019; 13, 1152. [DOI: https://dx.doi.org/10.3389/fnins.2019.01152] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31736693][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6828644]

24. Zhang, S et al. Comparing current steering technologies for directional deep brain stimulation using a computational model that incorporates heterogeneous tissue properties. Neuromodulation: J Int Neuromodulation Soc; 2020; 23, pp. 469-477.1:CAS:528:DC%2BB38XivFCqsbnN [DOI: https://dx.doi.org/10.1111/ner.13031]

25. Frank, JA; Antonini, M-J; Anikeeva, P. Next-generation interfaces for studying neural function. Nat. Biotechnol.; 2019; 37, pp. 1013-1023.1:CAS:528:DC%2BC1MXhsFKhtbjI [DOI: https://dx.doi.org/10.1038/s41587-019-0198-8] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31406326][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7243676]

26. Beauchamp, M. S. et al. Dynamic stimulation of visual cortex produces form vision in sighted and blind humans. English. Cell 181. Publisher: Elsevier, 774–783.e5. ISSN: 0092-8674, 1097-4172. https://www.cell.com/cell/abstract/S0092-8674(20)30496-7(2020).

27. Masuda, H; Shirozu, H; Ito, Y; Fukuda, M; Fujii, Y. Surgical Strategy for Directional Deep Brain Stimulation. Neurol. Med. Chir.; 2022; 62, pp. 1-12. [DOI: https://dx.doi.org/10.2176/nmc.ra.2021-0214] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8754682/

28. Contarino, M. F. et al. Directional steering. Neurology 83. Publisher: Wolters Kluwer, 1163–1169. https://doi.org/10.1212/WNL.0000000000000823 (2014).

29. Spencer, T. C. et al. Electrical Field Shaping Techniques in a Feline Model of Retinal Degeneration. eng. Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference 2018, 1222–1225. ISSN: 2694-0604 (2018).

30. Falcone, J. D. & Bhatti, P. T. Current steering and current focusing with a high-density intracochlear electrode array. eng. In: Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference 2011, 1049–1052. ISSN: 2694-0604 (2011).

31. Butson, CR; McIntyre, CC. Current steering to control the volume of tissue activated during deep brain stimulation. Brain Stimul.; 2008; 1, pp. 7-15. [DOI: https://dx.doi.org/10.1016/j.brs.2007.08.004] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19142235][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2621081]https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2621081/

32. McIntyre, CC; Grill, WM. Selective microstimulation of central nervous system neurons. Ann. Biomed. Eng.; 2000; 28, pp. 219-233.1:STN:280:DC%2BD3c3ksVKksQ%3D%3D [DOI: https://dx.doi.org/10.1114/1.262] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10784087]

33. Graziano, M. S. A., Taylor, C. S. R. & Moore, T. Complex movements evoked by microstimulation of precentral cortex. English. Neuron 34. Publisher: Elsevier, 841–851. ISSN: 0896-6273. https://www.cell.com/neuron/abstract/S0896-6273(02)00698-0 (2002).

34. Millar, J; Barnett, TG; Trout, SJ. The neurodyne: A simple mains-powered constant-current stimulus isolator. J. Neurosci. Methods; 1994; 55, pp. 53-57.1:STN:280:DyaK2M7pvVCgsQ%3D%3D [DOI: https://dx.doi.org/10.1016/0165-0270(94)90040-X] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/7891462]

35. Scholvin, J et al. Close-packed silicon microelectrodes for scalable spatially oversampled neural recording. IEEE Trans. Biomed. Eng.; 2016; 63, pp. 120-130. [DOI: https://dx.doi.org/10.1109/TBME.2015.2406113] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26699649][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4692190]https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4692190/

36. Saleh, M. S. et al. CMU Array: A 3D nanoprinted, fully customizable high-density microelectrode array platform. Science Advances 8. Publisher: American association for the advancement of science, eabj4853. https://doi.org/10.1126/sciadv.abj4853 (2022).

37. Wang, L. et al. High-density implantable neural electrodes and chips for massive neural recordings. en. Brain-X 2, e65. https://doi.org/10.1002/brx2.65 (2024).

38. Zhao, Z. et al. Ultraflexible electrode arrays for months-long high-density electrophysiological mapping of thousands of neurons in rodents. en. Nature Biomedical Engineering 7. Publisher: Nature Publishing Group, 520–532. ISSN: 2157-846X. https://www.nature.com/articles/s41551-022-00941-y (2023).

39. Wise, K. D., Angell, J. B. & Starr, A. An integrated-circuit approach to extracellular microelectrodes. IEEE Transactions on Biomedical Engineering BME-17. Conference Name: IEEE Transactions on Biomedical Engineering, 238–247. ISSN: 1558-2531. https://ieeexplore.ieee.org/document/4502738 (1970).

40. Wang, P. et al. Direct-Print 3D electrodes for large-scale, high-density, and customizable neural interfaces. en. Advanced Science n/a. 2408602. ISSN: 2198-3844. https://doi.org/10.1002/advs.202408602 (2025).

41. Lambrecht, J. M. et al. A distributed, high-channel-count, implanted bidirectional system for restoration of somatosensation and myoelectric control. eng. Journal of Neural Engineering 21. ISSN: 1741-2552 (2024).

42. Peterson, NR; Pisoni, DB; Miyamoto, RT. Cochlear implants and spoken language processing abilities: Review and assessment of the literature. eng. Restor. Neurol. Neurosci.; 2010; 28, pp. 237-250. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20404411][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2947146]

43. Firszt, J. B., Koch, D. B., Downing, M. & Litvak, L. Current steering creates additional pitch percepts in adult cochlear implant recipients. eng. Otology & Neurotology: Official publication of the american otological society, american neurotology society [and] European Academy of Otology and Neurotology 28, 629–636. (2007).

44. Wang, C et al. Vagus nerve stimulation: A physical therapy with promising potential for central nervous system disorders. eng. Front. Neurol.; 2024; 15, 1516242. [DOI: https://dx.doi.org/10.3389/fneur.2024.1516242] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39734634][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11671402]

45. Mirzakhalili, E., Barra, B., Capogrosso, M. & Lempka, S. F. Biophysics of temporal interference stimulation. English. Cell Systems 11. Publisher: Elsevier, 557–572.e5. ISSN: 2405-4712, 2405-4720. https://www.cell.com/cell-systems/abstract/S2405-4712(20)30372-0 (2020).

46. Jabban, L. et al. The comfort of temporal interference stimulation on the forearm: computational and psychophysical evaluation. en. Journal of Neural Engineering 22. Publisher: IOP Publishing, 026044. ISSN: 1741-2552. https://doi.org/10.1088/1741-2552/adc33b (2025).

47. Janson, AP; Anderson, DN; Butson, CR. Activation robustness with directional leads and multi-lead configurations in deep brain stimulation. eng. J. Neural Eng.; 2020; 17, 2020JNEng.17b6012J [DOI: https://dx.doi.org/10.1088/1741-2552/ab7b1d] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32116233][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7405888]026012.

48. Chaturvedi, A; Foutz, TJ; McIntyre, CC. Current steering to activate targeted neural pathways during deep brain stimulation of the subthalamic region eng. Brain Stimul.; 2012; 5, pp. 369-377. [DOI: https://dx.doi.org/10.1016/j.brs.2011.05.002] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22277548]

49. Bucksot, J. E. et al. Validation of a parameterized, open-source model of nerve stimulation. eng. Journal of Neural Engineering 18. ISSN: 1741-2552 (2021).

50. Brinke, TRT et al. Directional versus ring-mode deep brain stimulation for Parkinson’s disease: Protocol of a multi-centre double-blind randomised crossover trial eng. BMC Neurol.; 2023; 23, 372. [DOI: https://dx.doi.org/10.1186/s12883-023-03387-0] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37853327][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10583384]

51. Mishra, A; Unadkat, P; McBriar, JD; Schulder, M; Ramdhani, RA. An institutional experience of directional deep brain stimulation and a review of the literature. Neuromodulation: Technol Neural Interface; 2024; 27, pp. 544-550. [DOI: https://dx.doi.org/10.1016/j.neurom.2022.12.008] https://www.sciencedirect.com/science/article/pii/S1094715922014088

52. Udupa, K; Chen, R. The mechanisms of action of deep brain stimulation and ideas for the future development. Prog. Neurobiol.; 2015; 133, pp. 27-49. [DOI: https://dx.doi.org/10.1016/j.pneurobio.2015.08.001] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26296674]https://www.sciencedirect.com/science/article/pii/S030100821500088X

53. Kunigk, N. G., Urdaneta, M. E., Malone, I. G., Delgado, F. & Otto, K. J. Reducing behavioral detection thresholds per electrode via synchronous, spatially-dependent intracortical microstimulation. English. Frontiers in Neuroscience 16. Publisher: Frontiers. ISSN: 1662-453X. https://doi.org/10.3389/fnins.2022.876142/full (2022).

54. Slopsema, JP et al. Orientation-selective and directional deep brain stimulation in swine assessed by functional MRI at 3T eng. Neuroimage; 2021; 224, [DOI: https://dx.doi.org/10.1016/j.neuroimage.2020.117357] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32916285]117357.

55. Grill, WM. Temporal pattern of electrical stimulation is a new dimension of therapeutic innovation. Curr Opin Biomed Eng; 2018; 8, pp. 1-6. [DOI: https://dx.doi.org/10.1016/j.cobme.2018.08.007] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30906909][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6426311]https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6426311/

56. Hess, CW; Vaillancourt, DE; Okun, MS. The temporal pattern of stimulation may be important to the mechanism of deep brain stimulation. Exp. Neurol.; 2013; 247, pp. 296-302. [DOI: https://dx.doi.org/10.1016/j.expneurol.2013.02.001] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23399890][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3742615]https://www.sciencedirect.com/science/article/pii/S0014488613000447

57. Dorval, A. D., Kuncel, A. M., Birdno, M. J., Turner, D. A. & Grill, W. M. Deep brain stimulation alleviates parkinsonian bradykinesia by regularizing pallidal activity. Journal of Neurophysiology 104. Publisher: American Physiological Society, 911–921. ISSN: 0022-3077. https://doi.org/10.1152/jn.00103.2010 (2010).

58. Okorokova, E. V., He, Q. & Bensmaia, S. J. Biomimetic encoding model for restoring touch in bionic hands through a nerve interface. en. Journal of Neural Engineering 15. Publisher: IOP Publishing, 066033. ISSN: 1741-2552. https://doi.org/10.1088/1741-2552/aae398 (2018).

59. Soleimani, G. et al. Closing the loop between brain and electrical stimulation: Towards precision neuromodulation treatments. en. Translational Psychiatry 13. Publisher: Nature Publishing Group, 1–17. ISSN: 2158-3188. https://www.nature.com/articles/s41398-023-02565-5 (2023).

60. Guidotti, R. et al. When neuromodulation met control theory. eng. Journal of Neural Engineering. ISSN: 1741-2552 (2024).

61. Jun, J. J. et al. Fully integrated silicon probes for high-density recording of neural activity. Nature 551. Publisher: Nature Publishing Group, 232–236. ISSN: 1476-4687 (2017).

62. Karl, JA; Joyce, J; Ouyang, B; Verhagen Metman, L. Long-term clinical experience with directional deep brain stimulation programming: A retrospective review en. Neurol Therapy; 2022; 11, pp. 1309-1318. [DOI: https://dx.doi.org/10.1007/s40120-022-00381-5]

63. Schüpbach, WMM et al. Directional leads for deep brain stimulation: Opportunities and challenges en. Mov. Disord.; 2017; 32, pp. 1371-1375. [DOI: https://dx.doi.org/10.1002/mds.27096] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28843016]