Noninvasive Temporal Interference Electrical Stimulation for Spinal Cord Rehabilitation

Spinal cord injury (SCI) is a debilitating central nervous system disorder that can result in the permanent loss of motor, sensory, and autonomic functions below the level of injury^1,2. Consequently, the treatment and rehabilitation of SCI patients have become a focal point of both scientific research and clinical practice. Traditional treatment approaches, including pharmacological and physical therapies, have certain limitations in promoting functional recovery^3,4,5,6. Among physical therapies, spinal cord electrical stimulation has emerged as an effective strategy for SCI rehabilitation, which can be categorized into invasive and noninvasive modalities^7,8. Invasive spinal cord electrical stimulation, such as epidural spinal cord stimulation (eSCI), delivers direct electrical stimulation via implanted electrodes but carries risks of infection and scar tissue formation^9,10. In contrast, noninvasive techniques, such as transcutaneous electrical nerve stimulation (TENS), are limited in their ability to effectively reach deep spinal structures, thereby compromising therapeutic efficacy^11,12.

Temporal interference (TI) stimulation is an emerging neuromodulation technology that enables noninvasive stimulation of deep tissues through a specific mode of electrical current delivery^13,14. This technique involves placing two pairs of electrodes on the skin surface to deliver electrical currents at slightly different kilohertz frequencies. Based on the principle of interference, this setup generates a unique low-frequency envelope (ranging from a few hertz to several tens of hertz) within deep tissues, thereby enabling targeted neuromodulation. This distinct working mechanism allows TI stimulation to overcome the depth limitations of conventional neuromodulation techniques, providing an effective intervention for deep neural structures without invasive procedures. Unlike TENS, TI achieves deeper penetration with high spatial specificity, and unlike eSCI, it avoids surgical risks, offering a safer, more accessible alternative for SCI neuromodulation. TI stimulation has been investigated for the treatment of various diseases, such as movement disorders and depression. In incomplete SCI, as some neural pathways remain intact, TI stimulation is highly likely to enhance the activity of remaining neural circuits, thereby promoting neuroplasticity and functional recovery^15,16. Thus, TI stimulation holds significant promise as a neuromodulation strategy for SCI treatment¹⁷.

However, current TI stimulation hardware systems are primarily designed for transcranial applications, and there is a lack of TI systems specifically developed for spinal cord stimulation. Due to anatomical and electrophysiological differences between the head and the torso, existing TI stimulation devices designed for the head are not fully applicable to spinal stimulation, leading to challenges in output parameter optimization and electrode placement. When performing TI stimulation on the head, a fixed leadfield coordinate system (such as the 10-10 system) is often used to facilitate electrode positioning on the head. However, this system is not applicable to the torso. Furthermore, because TI stimulation generates low-frequency envelopes deep within biological tissues, it is difficult to predict the resulting electric field distribution based solely on manual electrode placement. Instead, computational simulations are typically required to visualize and optimize the internal electric field distribution. At present, however, there is no established workflow for electric field simulation and parameter optimization for spinal TI stimulation, which poses significant challenges for its clinical application. Parameters such as electrode placement, stimulation frequencies, and current amplitude directly influence the electric field distribution and the amplitude of the low-frequency envelope, modulating neural activity and promoting neuroplasticity^13,17.

The objective of this study is to develop a convenient and effective workflow for TI electric field simulation and parameter optimization, along with a TI hardware system tailored for spinal cord injury treatment. Through electric field simulation and parameter optimization, we aim to determine electrode placement configurations that maximize the envelope field amplitude of TI at specific SCI target regions, thereby enhancing therapeutic efficacy. Additionally, to facilitate the practical implementation of optimized electrode configurations, we have designed a new electrode coordinate positioning method for spinal cord TI stimulation based on the original TI hardware system for the head. This system is intended to simplify electrode positioning and improve operational feasibility in clinical settings.

This study involves human subjects and was conducted in accordance with the Declaration of Helsinki. Ethical approval was obtained from the Institutional Review Board of Zhejiang University. Written informed consent was obtained from all participants prior to their inclusion, ensuring they were fully informed of the study's purpose, procedures, potential risks, and their right to withdraw at any time without penalty. The reagents and the equipment used in this study are listed in the Table of Materials.

Contraindications and special considerations
SCI patients are assessed for eligibility using a medical history questionnaire and physical examination to identify conditions affecting participation:
Inclusion criteria: (1) Age between 18 and 80 years (male or female); (2) Incomplete SCI graded as ASIA B, C, or D, with onset of 1-6 months; (3) No changes in ASIA assessment within the past week; (4) Stable medication regimen throughout the study period; (5) Willingness to comply with all study requirements, including participation in all required training sessions and rehabilitation assessments.
Exclusion criteria¹⁷: (1) Motor function limitations due to neurological disorders (e.g., stroke, multiple sclerosis, traumatic brain injury); (2) Presence of any unstable or severe medical conditions (e.g., uncontrolled hypertension, heart failure); (3) History of epilepsy; (4) Contraindications to electrical stimulation (e.g., implanted electronic devices, pacemakers, metallic implants).

1. Materials

1. Verify the completeness of required materials (see Table of Materials).
2. Evaluate SCI patients using magnetic resonance imaging (MRI) or computed tomography (CT) scans.
3. Open electromagnetic simulation software to perform electric field simulations (all steps are implemented within the software, eliminating the need to switch to other software or use additional scripts). Before conducting electric field simulations, ensure that the electromagnetic simulation software contains human models (male or female).
   1. If these models are not available, select Model/Phantom - Humans in the ribbon, then download the human model. Once the download is complete, import the model into the current project for subsequent simulations.
4. Press the power button to turn on the TI stimulator before each stimulation session, then connect the electrode adapter using the designated connection cables (Figure 1). Check the battery status on the device screen. Charge the device immediately if the battery level is below 20%, using the designated power adapter plugged into a standard electrical outlet.
   NOTE: Incompatible adapters may damage the battery or cause irregularities in the TI stimulator's output.
5. Inspect each Ag/AgCl electrode visually for cracks, discoloration, or residue. Clean the electrodes with isopropyl alcohol and a soft cloth to remove contaminants. Then apply a thin and even 1 mm layer of conductive gel to the electrode surface to reduce contact impedance and prevent skin irritation during stimulation.
   1. Additionally, ensure to have precise information regarding the dimensions of the electrodes, including their diameter and thickness, as these parameters play a critical role in subsequent electric field simulations.
6. Connect the two pairs of electrode cables to the electrode adapter by matching the cable connectors to the labeled ports.

2. Electric field simulation and parameter optimization

NOTE: The overall workflow of the electric field simulation consists of three main steps: constructing the geometric model (including the human model and electrodes), defining simulation conditions (material properties, boundary conditions, and mesh generation), and finally performing computations to visualize the electric field distribution in the target region of spinal cord (Figure 2). Parameter optimization involves simulating electric fields for various candidate electrode pair configurations, calculating the average electric field intensity in the target region, and identifying the configuration that maximizes this intensity. The specific steps are as follows:

1. Launch the electromagnetic simulation software on a computer with at least 16 GB of RAM and a multi-core processor. Create a new project by selecting File - New Project in the ribbon.
2. Geometric model construction
   1. Click on the Model tab in the ribbon and import a human model by selecting Model/Phantom - Humans and choosing a male or female static model (Figure 3).
   2. Identify the vertebral location closest to the SCI site using anatomical landmarks (e.g., C5 is approximately 2-3 cm below the C7 spinous process, palpable as a prominent bony protrusion at the base of the neck). Set the coordinate of the skin surface directly above the central region of the selected vertebra as the origin (Sketch - Point). Define the horizontal and vertical axes as the x- and y-axes, respectively (Figure 4).
      NOTE: In the demonstration of the protocol, we select the spinal cord region wrapped by the C5 vertebra as the stimulation target. This is not a fixed choice, but it was made for the convenience of demonstration. When applying the protocol in practice, any specified spinal cord target area can be arbitrarily selected.
   3. For each electrode pair configuration, place the four electrodes symmetrically about the origin (Figure 4), with one pair on each side of the spine. Define positions by the horizontal distance (d₁) and vertical distance (d₂) from the origin, such that the coordinates of the four electrodes in each configuration are (d₁, d₂), (d₁, -d₂), (-d₁, d₂), and (-d₁, -d₂).
      NOTE: In this demonstration, 25 electrode pair configurations are evaluated, with adjacent electrodes in each quadrant spaced 1 mm apart horizontally and 5 mm apart vertically, which allows the candidate electrodes to be distributed within a suitable range relative to the target area. The specific electrode placement procedure is as follows. Using the first quadrant as an example:
      1. Create a cylindrical electrode by clicking on Solids - Cylinder in the ribbon with dimensions corresponding to the actual electrodes used in the clinical treatment. The thickness may be slightly increased to account for minor intersections with the skin surface in the simulation model.
         NOTE: In this demonstration, the electrode diameter was set to 10 mm, and the thickness to 5 mm.
      2. Click on Move in the ribbon to place the cylinder on the posterior skin surface with slight embedding. Adjust its center to (10 mm, 10 mm), representing a horizontal and vertical distance of 10 mm from the origin.
      3. Duplicate the cylinder in the Model explorer and shift it rightward by 11 mm to position the second electrode at (21 mm, 10 mm). Repeat this process to create additional electrodes at (32 mm, 10 mm), (43 mm, 10 mm), and (54 mm, 10 mm).
      4. Copy the first electrode and shift it upward by 15 mm to position the second electrode at (10 mm, 25 mm). Repeat this process to generate electrodes at (10 mm, 40 mm), (10 mm, 55 mm), and (10 mm, 70 mm).
      5. Repeat this process, generate 25 electrodes (5 rows × 5 columns) within the first quadrant.
         NOTE: Each TI stimulation configuration uses four electrodes (two pairs) to deliver two kilohertz-frequency currents that interfere to create a low-frequency envelope in the spinal cord. To optimize placement, 25 configurations, each with four electrodes, are tested using a 5 x 5 grid of positions per quadrant, totaling 100 electrodes across all simulations. The size of the electrodes and the distance between electrodes in different groups are specified according to actual conditions, and this demonstration only provides an example.
   4. For the placement procedure of the second, third, and fourth quadrants, follow the same process. Upon completion, all 25 electrode pair configurations (a total of 100 electrodes) are arranged as shown in Figure 4.
   5. Rename electrodes systematically by right-clicking on the corresponding electrode in the Model explorer (e.g., "ROn" for first quadrant, "LOn" for second, "LBn" for third, "RBn" for fourth, where n is the group number from 1 to 25).
   6. Inspect electrode placement visually in the 3D model view. Adjust positions using the Move to ensure each electrode embeds slightly (0.5 mm) into the skin surface without gaps.
3. Simulation conditions setting
   1. Click on the Simulation tab and create a new simulation by clicking on New - EM LF Electro - Ohmic Quasi-Stat in the ribbon. Right click and rename it "LF-R1". Since TI stimulation operates at low frequencies, use "Ohmic Quasi-Stat" solver to consider only the quasi-static approximation.
   2. Click on the Setup panel under Simulation explorer, set the frequency to 1040 Hz.
   3. Drag the human model from the Multi-Tree into LF-R1 - Materials, which will automatically assign tissue-specific conductivity values at 1040 Hz (Table 1).
   4. In Boundary Conditions, modify the default Boundary settings to Flux by selecting Boundary Type and choosing Flux from the dropdown menu. Then, create two new "Boundary settings" by clicking on new setting and assign input voltages of +1 V and 0 V ( Figure 5).
      1. Assign the electrodes "RO1" and "RB1" to these settings by dragging them into the respective "Boundary Settings", respectively, indicating that "RO1" outputs +1 V while "RB1" serves as the return path.
   5. Retain the default settings for "Sensors" and adjust the Grid by setting Maximum Step to 1 mm and Geometry resolution to 1 mm, based on computational resources.
   6. Create a new simulation "LF-L1" by duplicating "LF-R1", set the frequency to 1000 Hz, and assign the electrodes "LB1" and "LO1" to the +1 V and 0 V "Boundary settings".
      NOTE: This setup mirrors "LF-R1" but with the opposite electrode configuration, generating the interference effect. All other settings remain unchanged. This completes the first electrode pair simulation setup .
   7. Duplicate LF-R1 and LF-L1, rename them as LF-R2 and LF-L2, and modify the Boundary settings to assign the corresponding electrodes. Remove prior electrodes from "Grid" and "Voxels" and add the new set for each pair. Repeat for all 25 groups.
   8. Once all simulations are configured, select all simulations in the Simulation tab, click on Auto Grid Update in the ribbon (Click on the Task Manager in the lower right corner. If the window indicates that the grid has been constructed, this confirms that the steps are proceeding normally), and then select Run - Batch Run to run all simulations simultaneously.
      NOTE: The frequencies of 1040 Hz (LF-R1) and 1000 Hz (LF-L1) are chosen to produce a low-frequency envelope of 40 Hz (1040 Hz - 1000 Hz) through interference, which is the effective stimulation frequency in the target spinal cord region for neuromodulation¹⁷.
4. Perform statistical analysis
   1. Click on the Analysis tab, select LF-R1 - Overall Field - Sensor Extractor, and click on Overall Field to generate an Overall Field .
   2. Click on Overall Field - EM E(x,y,z,f0) - Field Data Tools - Field Scaling to convert the electric field distribution from an input voltage of 1V to an input current of 1mA. The scale factor is determined as follows (Figure 6):
      1. In the Model tab, create a cubic volume (Block RO1) enclosing electrode RO1 by selecting Solids - Block in the ribbon and setting dimensions to fully encompass the electrode (e.g., 12 mm × 12 mm × 7 mm).
      2. Drag Block RO1 from the Multi-tree into the Analysis panel, generating two identical "Block RO1" modules.
      3. Select Overall Field under LF-R1 and the first Block RO1 in the Model explorer, then click on Surface and EM E(x,y,z,f0) simultaneously. Click on Flux Evaluator - List Viewer to display the "Total Flux" value. Compute the scale factor by dividing 0.001 by the Total Flux value.
   3. Repeat the same process for "LF-L1" .
   4. Multi-select LF-R1 and LF-L1 under Field Scaling in the Analysis explorer, then click on Max Modulation in the ribbon to couple the electric field distributions from the two electrode pairs. Set "Weight A" and "Weight B" to 2 to reflect 2 mA output per electrode pair (Figure 7).
   5. Extract the average electric field intensity in the target region of the spinal cord:
      1. Apply a "Mask Filter" by selecting Field Data Tools - Mask Filter in the ribbon to retain only the "Spinal Cord" region.
      2. Click on LF-R1 in the Analysis explorer and click on Field Data Tools - Crop in the ribbon to isolate the target spinal cord region (Figure 8).
      3. Click on Statistics - Table Viewer in the ribbon to display the average electric field intensity for the cropped region (Normal value range: 0.1-2 V/m).
   6. Repeat the process for all 25 electrode pair configurations and acquire the average electric field intensity at the target of all 25 groups .
   7. Compare the average electric field intensity for each configuration and identify the configuration with the highest intensity as the optimal setup.

3. Electrode positioning and device setup

1. Ask the SCI patient to sit comfortably on a chair with a backrest, ensuring their torso is upright and back exposed by removing or adjusting clothing.
2. Identify the vertebral level closest to the SCI site by palpating the spinous processes along the spine. Locate the C7 spinous process (prominent at the base of the neck) and count downward to the target vertebra (e.g., C5 is two vertebrae above C7). Mark the skin surface above the target vertebra's spinous process with a washable marker (58-7726, Crayola Inc.) to designate the origin¹⁸.
3. Based on the previously optimized electrode placement parameters (d₁, d₂), measure the four electrode positions using a flexible measuring tape. From the origin, measure d₁ horizontally (left and right) and d₂ vertically (up and down) to locate positions in the four quadrants: (d₁, d₂), (d₁, -d₂), (-d₁, d₂), (-d₁, -d₂). Mark each position with a washable marker.
4. Clean the skin at the marked electrode sites with alcohol wipes to remove oils and debris. Inspect the skin to ensure it is intact, with no cuts or abrasions. Apply 1.5 mL of conductive gel to each site using a syringe or applicator, spreading it evenly over a 10 mm diameter area.
5. Minimize tension on the electrode cables by securing excess cable length with tape. Suspend cables to hang naturally by attaching them to the chair or a nearby stand, reducing the risk of accidental detachment.
6. Attach the electrodes to the marked positions on the patient's back by pressing each Ag/AgCl electrode firmly onto the gel-covered skin. Secure each electrode with adhesive tape, covering the edges to prevent movement.
   NOTE: The right-side and left-side electrodes each form an electrode pair. Designate the right-side upper electrode as the anode and the lower electrode as the cathode. Designate the left-side upper electrode as the cathode and the lower electrode as the anode, forming a mirror-image configuration.
7. Once the electrodes are properly positioned and all connections are verified, power on the TI stimulator by pressing the power button. Click on Global Parameter Settings and configure the following stimulation parameters: Stimulation mode: tTIS, Stimulation type: Standard Stimulation, Total stimulation duration: 1200 s, Stimulation waveform: Sine wave by selecting the corresponding options in the menu.
8. Perform impedance measurement for the two selected stimulation channels. Click on Start Impedance Test to ensure that the impedance falls within the acceptable range (the stimulator will automatically stop output if the load impedance exceeds 12 kΩ). The optimal impedance range for effective stimulation is 6-8 kΩ.
9. Configure the stimulation frequency and output current amplitude for each channel. Channel 1: Frequency set to 1040 Hz, current amplitude set to 2 mA. Channel 2: Frequency set to 1000 Hz, current amplitude set to 2 mA¹⁵.

4. Stimulation

1. Ensure that the participant remains in a comfortable seated position throughout the procedure and remains fully conscious at all times.
2. Activate the stimulator by pressing the Start switch on the device's interface to initiate stimulation.
3. Monitor the patient's feedback during stimulation by asking them to report any sensations (e.g., tingling, discomfort). Reduce the current amplitude by 0.5 mA increments or pause stimulation immediately by pressing the Stop button if the patient reports significant discomfort or a strong stinging sensation.
   NOTE: Expect a mild tingling or itching sensation, which is a normal physiological response to transcutaneous electrical stimulation, as observed in this study and in prior research on similar techniques¹⁷. This sensation typically subsides within 2-3 min.

5. Postprocedural steps

1. After the stimulation session is completed, measure the impedance status again by selecting Start Impedance Test to assess changes post-stimulation. Record the values for comparison.
2. Verify the patient's physical stability by checking their balance and comfort while seated. Remove medical tape gently by peeling it slowly from the skin to avoid irritation. Detach electrodes carefully by lifting them from one edge.
3. Rinse electrodes with clean water under a faucet and air-dry them on a clean towel. Wipe the patient's skin at electrode sites with a damp paper towel to remove residual gel.
4. Administer a standardized questionnaire to the patient, asking about sensory changes (e.g., tingling, numbness), motor improvements, and adverse events (e.g., skin irritation, pain). Record responses for analysis¹⁹.

When conducting TI simulations without errors, the average electric field intensity in the target spinal cord region stimulated by the current group of electrode pairs can be obtained. Taking Group 10 stimulating the C5 target area as an example (Figure 9), the "Volume Weighted Average" displayed in the interface is 0.50 V/m. Additionally, by clicking "Max Modulation - Mask Filter - Viewers - Surface Viewer", a 3D view of the electric field distribution on the spinal cord can be preserved while setting other tissues to semi-transparent. This allows for an intuitive observation of the electric field distribution of Group 10 around the C5 target area (Figure 10).

After completing simulations for all groups, the average electric field intensity at each target area is analyzed and compared. For instance, in simulations conducted on the model, TI stimulation was applied to three target areas: C5, T7, and L3 (Figure 11), as reported by Xie et al.20. The results indicate that a smaller d2 results in a lower average electric field intensity in the target region. The optimal (d1, d2) values for the three target areas were found to be (32 mm, 70 mm) for C5, (10 mm, 40 mm) for T7, and (10 mm, 70 mm) for L3.

In practice, when TI stimulation is first applied, a mild itching or slight tingling sensation may occur. This is a normal physiological response, indicating that current is passing through the skin, as observed in this study and supported by studies of similar electrical stimulation techniques19. The sensation typically diminishes within a few minutes.

At present, clinical applications of TI stimulation for SCI remain limited, and its therapeutic efficacy requires further validation. However, existing clinical studies have demonstrated that two weeks of continuous TI stimulation leads to significant improvements in neurological function, motor strength, sensory perception, and functional independence in SCI patients (Table 2), as reported by Cheng et al.17. These findings support the hypothesis that TI stimulation is an effective therapeutic approach for SCI treatment.

Figure 1: Electrode placement during clinical treatment based on electric field simulation.Two pairs of electrodes were placed according to the optimal configuration determined through electric field simulation and parameter optimization. The stimulation target (e.g., C5) was identified, and the point on the skin directly above this target-perpendicular to the skin surface-was defined as the origin. Using the optimized coordinates (d1, d2) relative to the origin, the placement positions of the two electrode pairs were determined. Please click here to view a larger version of this figure.

Figure 2: The pipeline of electric field simulation and parameter optimization. A total of 25 candidate groups are evaluated, with each group consisting of two electrode pairs: one pair positioned on the right side of the target region (R2 Pair) and the other on the left side (L2 Pair). The four electrodes in each group are placed at an identical horizontal distance (d1) and vertical distance (d2) from the origin, allowing each group to be represented as (d1, d2). By systematically positioning the electrode pairs and setting up the simulation conditions, the average electric field intensity within the target region is calculated for all groups. The groups are then compared, and the Best group (d1, d2) is determined based on the highest average electric field intensity. Please click here to view a larger version of this figure.

Figure 3: Human model used for simulation. The Duke V3.0 Static human model was selected and imported into through the "Model/Phantom" option in the ribbon interface. This model was downloaded and incorporated for use in the simulation environment. Please click here to view a larger version of this figure.

Figure 4: Electrode placement in simulation and parameter optimization. Two pairs of electrodes were placed in each simulation. All electrode configurations used during parameter optimization are also shown. Please click here to view a larger version of this figure.

Figure 5: Boundary settings for the LF-R1 simulation. The boundary conditions for the LF-R1 simulation were configured by first selecting "Boundary Settings" in the software. In the "Controller" panel, the "Boundary Type" was set to "Flux". Two "Boundary Settings - Dirichlet" entries were then created by right-clicking on "Boundary Conditions" in the Explorer and selecting "New Settings". In the "Multi-tree", the anode and cathode of one electrode pair were assigned to the respective Dirichlet boundary settings. The "Constant Potential" was set to 1 V for the anode and 0 V for the cathode in the Controller panel. Please click here to view a larger version of this figure.

Figure 6: Conversion of the electric field distribution from a 1 V input to a 1 mA input. To convert the electric field distribution obtained using a 1 V input to that corresponding to a 1 mA input, a cubic volume (Block RO1) was created around electrode RO1 in the Model tab by selecting "Solids - Block" from the ribbon and adjusting the dimensions (e.g., 12 mm × 12 mm × 7 mm) to fully encompass the electrode. The "Block RO1" object was then dragged from the "Multi-tree" into the "Analysis" panel, generating two identical modules. Within the "Model" explorer, "Overall Field" under "LF-R1" and the first instance of "Block RO1" were selected, followed by activation of the "Surface" and "EM E(x,y,z,f0)" options. The "Flux Evaluator - List Viewer" was used to display the "Total Flux" value. The scale factor was determined by dividing 0.001 by the Total Flux value. Please click here to view a larger version of this figure.

Figure 7: Electric field modulation and envelope amplitude calculation. The electric fields generated by the two electrode pairs in one group were modulated, and their envelope amplitudes were calculated. The "LF-R1" and "LF-L1" entries under "Field Scaling" in the "Analysis" explorer were selected together, and the "Max Modulation" function in the ribbon was used to couple the electric field distributions from the two electrode pairs. The parameters "Weight A" and "Weight B" were both set to 2, corresponding to an output of 2 mA per electrode pair. Please click here to view a larger version of this figure.

Figure 8: Isolation of the spinal cord target region and calculation of average electric field intensity. The target spinal cord region was cropped and extracted to evaluate the electric field intensity. In the "Analysis" explorer, the "LF-R1" field was selected, and the "Field Data Tools - Crop" function in the ribbon was used to isolate the desired area. The average electric field intensity within this region was subsequently calculated. Please click here to view a larger version of this figure.

Figure 9: Average electric field intensity at target of spinal cord in TI simulation (Group 10). Please click here to view a larger version of this figure.

Figure 10: 3D view of electric field distribution of spinal cord in TI simulation (Group 10). Please click here to view a larger version of this figure.

Figure 11: Average electric field intensity at the target of spinal cord simulated using 25 groups. The optimal (d1, d2) values for the three target areas were found to be (32 mm, 70 mm) for C5, (10 mm, 40 mm) for T7, and (10 mm, 70 mm) for L3. This figure is modified from Xie et al.20. Please click here to view a larger version of this figure.

Table 1: The electrical conductivities of relative tissues at 1 kHz. Please click here to download this Table.

Table 2: Demographic and clinical characteristics of the participants stimulated by TI. Modified from Cheng et al.17. Please click here to download this Table.

Critical steps

Setting up simulation conditions
When electrodes are placed on the surface of the human model's skin, the cylindrical electrodes are partially embedded into the skin to ensure there is no air gap between the electrodes and the skin. Otherwise, the current cannot pass through the air and into the human body. The distance from the electrode to the origin (d₁, d₂) is measured along the skin surface, as the skin is not perfectly flat. This distance may differ slightly from a direct linear measurement.

When material properties are assigned to the geometric models, no material properties are assigned to the electrodes since they serve as the source. When the grid is configured, the mesh resolution is adjusted according to the available computational resources. An overly fine mesh slows down computation, whereas an excessively coarse mesh may compromise the accuracy of the simulation. The "Padding Setting" is set to Manual, and both "Bottom Padding" and "Top Padding" are set to 0, as the low-frequency solver does not require computation in the peripheral regions. Before running the simulation, "Auto Grid Update" is selected and "Create Voxels" has been clicked; otherwise, the simulation cannot proceed.

When "Field Scaling" is performed, the created block region fully encloses the electrode. For each electrode pair, only one block is needed to encapsulate a single electrode. If the block encloses the anode, the resulting "Total Flux" is positive, whereas if it encloses the cathode, the "Total Flux" is negative. When the "Scaling Factor" is calculated, the "Total Flux" is always treated as a positive value, because the "Scaling Factor" is always a positive value. Additionally, if the "Total Flux" is extremely small or close to zero, verification is conducted to check whether the electrode pair corresponding to the "Scaling Factor" calculation matches the selected block. If any unexpected issues arise during the simulation, carefully review the setup conditions to identify potential errors.

Pre-checks before TI stimulation
Before initiating treatment, ensure that participants have no contraindications to TI stimulation, as specified in the protocol. In particular, it should be confirmed that the participant (1) has no relevant contraindications, (2) has no metallic implants in the body,(3) has intact skin at the electrode placement sites, and (4) is not pregnant.

Electrode wires should be inspected to ensure they are intact, as they are thin and prone to damage. Additionally, after repeated use, we observed that electrolytic deposits may accumulate on the surface of the Ag/AgCl electrodes, potentially increasing impedance during our procedures. If increased impedance is detected, replace the electrodes with new ones. The participant should be ensured to be in a comfortable and stable position, as maintaining stillness is crucial for the proper delivery of current.

Considerations during TI stimulation
The participant should remain awake throughout the stimulation session. The participant's feedback should be continuously monitored. If the participant reports severe pain, pause the stimulation or reduce the intensity promptly.

Post-treatment considerations
After each session, it is recommended to conduct a questionnaire survey to collect data on the safety and potential adverse effects of TIstimulation.

Usability and clinical implementation
The TI stimulation protocol offers several intensities that enhance its usability in clinical settings. Its noninvasive nature eliminates the risks associated with surgical implantation, such as infection or scar tissue formation, making it more acceptable to patients compared to epidural stimulation^9,10. The standardized simulation workflow, using electromagnetic simulation software, ensures precise electrode placement tailored to specific SCI regions, improving reproducibility across patients¹⁷. The protocol's setup, once optimized, requires approximately 15 min for simulation for each group and 15-20 min for electrode positioning and device configuration, which is feasible within a clinical session. Additionally, the potential to automate electrode placement using pre-configured simulation templates could further enhance setup efficiency, reducing preparation time in high-volume clinics.

However, clinical implementation faces several challenges that impact real-world feasibility. The simulation process requires high-performance computing resources (e.g., 16 GB RAM, multi-core processor), which may not be available in resource-limited settings, such as rural or underfunded clinics. Patient-specific anatomical variations, such as spinal cord depth or skin thickness, necessitate individualized simulations, increasing preparation time and complexity. For example, patients with obesity or severe SCI (e.g., ASIA A) may require additional adjustments to achieve the target electric field intensity (≥ 0.5 V/m) at certain hardware limitations (maximum output current of 2-5 mA), potentially reducing efficacy. The TI stimulator's battery life, typically 1-2 h, may also limit continuous use in busy clinical environments, requiring careful scheduling or backup power sources.

To improve feasibility, clinics could adopt cloud-based simulation platforms to reduce hardware demands or develop simplified electrode placement guides based on common SCI injury levels. These strategies would enhance accessibility and streamline setup, making the protocol more practical for diverse clinical settings. Despite these challenges, the protocol's ability to non-invasively target deep spinal structures positions it as a promising tool for SCI rehabilitation, particularly in well-equipped facilities with trained staff.

Possible modifications
In this demonstration, only vertebra C5 was selected as the target region of the spinal cord for simulation. However, in practical applications, the target region of the spinal cord can be freely chosen based on the specific spinal cord injury. For this TI stimulation simulation, a pair of electrodes was placed on each side of the origin. The right-side electrode pair was stimulated at 1040 Hz, while the left-side electrode pair was stimulated at 1000 Hz. For the right-side electrode pair, the upper electrode served as the anode, and the lower electrode served as the cathode. Conversely, for the left-side electrode pair, the polarity was reversed. In future explorations, parameters such as the positioning of electrode pairs, selection of anodes and cathodes, stimulation frequencies, and stimulation intensities can be flexibly adjusted to achieve optimal parameter configurations. During parameter optimization, the horizontal and vertical electrode spacing between different groups can be freely adjusted. Furthermore, in both simulation and actual TI stimulation, the number of electrode pairs used is not necessarily limited to two. MultipolarTI has become a significant area of ongoing research^21,22.

In clinical applications, in addition to measuring the horizontal distance (d₁) and vertical distance (d₂), the direct linear distance between each electrode and the origin, as well as the angle between this line and the horizontal axis, can also be measured to determine the precise electrode placement. Alternatively, rubber adhesive electrodes can be used as a substitute for Ag/AgCl electrodes. These electrodes can adhere directly to the skin on the back without the need for additional fixation measures.

The efficacy of TI stimulation for SCI rehabilitation depends on optimized simulation parameters, including frequency, current amplitude, and electrode placement, as highlighted by Rehman et al.²³in their review of trans-spinal stimulation (TSS). In our protocol, two frequencies (1000 Hz and 1040 Hz) generate a 40 Hz low-frequency envelope, chosen to enhance motor neuron excitability and promote neural plasticity, similar to TSS frequencies (20-50 Hz) that improve motor strength. Variations in envelope frequency could influence outcomes; for instance, lower frequencies (e.g., 20 Hz) may enhance reflex modulation, while higher frequencies (e.g., 100 Hz) could reduce spasticity, as seen in TSS studies. The current amplitude of 2 mA achieves an electric field intensity of 0.5-2 V/m, sufficient for neuromodulation, but higher amplitudes might increase motor responses, though with potential discomfort, while lower amplitudes could reduce efficacy. Future studies should explore a wider range of parameters to tailor TI stimulation for specific SCI outcomes, such as spasticity reduction or gait improvement.

Limitations
The human model used in the simulation is a standardized model rather than an individualized model, which may affect the accuracy of the simulation. Future research should further explore the construction of individualized torso simulation models. Additionally, in the simulation, the electrical properties assigned to each tissue were treated as constants, based on values reported in the literature^24,25,26,27. However, in reality, the distribution of electrical properties of tissues is more complex, which can be measured through some techniques^28,29,30.

The use of standardized models is generally acceptable for initial protocol development and clinical application, as they provide a reproducible framework for optimizing electrode placement across a broad patient population. However, their accuracy may be limited by inter-individual anatomical variations, such as spinal cord depth or tissue composition. To verify the accuracy of simulated TI fields in targeting specific spinal regions, clinical outcomes (e.g., motor or sensory improvements) can be correlated with predicted electric field distributions, as assessed through post-stimulation questionnaires or functional assessments.

Currently, there is still a lack of validation regarding the therapeutic efficacy and advantages of the results obtained through parameter optimization. Future clinical studies are required to validate the effectiveness of this approach.