Non-Invasive Brain Stimulation

• Non-invasive brain stimulation is a set of techniques that modulate brain activity using external electrical, magnetic, optical, and acoustic energies with high spatial and temporal precision.
• These methods leverage advanced simulation, deep-learning emulators, and optimization algorithms to target specific brain regions within strict safety and exposure limits.
• Recent advances include closed-loop protocols and robot-assisted systems that enhance reproducibility and personalization in clinical and experimental settings.

Non-invasive brain stimulation (NIBS) encompasses an array of physical techniques for modulating neural activity in the central nervous system without surgical access. The field unites methods based on electrical, magnetic, optical, and acoustic energy—each characterized by distinct mechanisms of action, spatial/temporal resolution, depth of targeting, and clinical/experimental relevance. NIBS protocols are widely deployed for causally probing circuit function, enhancing neurorehabilitation, ameliorating neuropsychiatric conditions, and exploring fundamental principles of network plasticity.

1. Physical Principles and Modalities

NIBS is implemented via external devices that generate focused or distributed fields to modulate neural excitability and plasticity. The principal approaches are:

• Transcranial Electric Stimulation (tES): Applies weak currents (<2–4 mA) via scalp electrodes, including direct (tDCS), alternating (tACS), random noise (tRNS), and temporal interference (tTIS) forms, producing subthreshold polarization or entrainment of cortical populations. Field propagation is governed by the quasi-static Laplace equation, $$\nabla \cdot [\sigma(\mathbf{r}) \nabla \Phi(\mathbf{r})] = 0,$$ with tissue-dependent conductivities (Tarazona et al., 2015).
• Transcranial Magnetic Stimulation (TMS): A rapidly changing current pulse in a wire coil generates time-varying magnetic fields; by Faraday’s law, the induced electric fields depolarize neurons primarily in superficial cortex. The induced field is $$\mathbf{E} = -\partial\mathbf{A}_0/\partial t - \nabla V,$$ with field magnitude and distribution dictated by coil geometry and subject anatomy (Gomez-Tames et al., 2020).
• Transcranial Focused Ultrasound (tFUS), Optically-generated Focused Ultrasound (OFUS), and Optically-generated Bessel Beam Ultrasound (OBUS): These approaches convert electromagnetic energy (laser pulses) into acoustic pressure via optoacoustic transducers for neural modulation at high spatial precision, including sub-100 μm lateral resolution and volumetric column targeting (Li et al., 2022, Li et al., 8 Jul 2025).
• Transcranial Photobiomodulation (tPBM): Delivers near-infrared light (600–1700 nm) through the scalp to drive mitochondrial photoreceptors (cytochrome c oxidase), modulating neuronal metabolism, ATP production, and oxidative stress with negligible thermal effect at clinical power densities (Ibrahimi et al., 2022, Li et al., 13 Jul 2024).

Magnetically induced methods (e.g., TMS) leverage rapid coil pulses to create tangential cortical E-fields, whereas electrical and optoacoustic modalities allow both distributed and highly focal targeting—including noninvasive access to deep brain regions through constructive interference or high-frequency focusing.

2. Biophysics, Targeting, and Field Modeling

The efficacy and specificity of NIBS are governed by the spatial distribution of induced electric (and in ultrasound, mechanical) fields, which depend on device configuration, anatomical heterogeneity, and computational modeling accuracy.

Electric and Magnetic Stimulation

• Conductivity profile and field spread: Realistic head models account for skull, scalp, CSF, and brain tissue, with the skull's resistivity dominating field attenuation and broadening, particularly for tES. Electric fields in the cortex typically reach 0.2–0.5 V/m at 1–2 mA tDCS (for 1 cm² electrodes), with spatial dispersion ≈ several cm for classic montages (Tarazona et al., 2015, Sheltraw et al., 2020).
• High-definition montages and optimization: Advances include multi-electrode tDCS arrays and convex optimization frameworks maximizing focality and directionality under safety constraints $$\omega(x)|A[I](x)| \le \epsilon, \quad \|I\|_{\mathcal{M}(\Gamma)} \leq 4\ \mathrm{mA},$$ with finite element solutions and alternating direction method of multipliers (ADMM) algorithms for rapid computation (Wagner et al., 2015).
• Deep brain targeting: Temporal interference (TI) and spatio-temporal Fourier synthesis (STFS) leverage superposed kHz-sinusoidal fields to produce amplitude-modulated low-frequency envelopes concentrated at depth. The beat envelope, $|E_\mathrm{AM}(\mathbf{r})|,$ is optimized via electrode geometry and currents, enabling focal activation (>1 V/m in a ~1.5 cm³ volume) while minimizing superficial field exposure (Zibandepour et al., 2023, Kish et al., 2 Apr 2024, Söderholm et al., 23 Jun 2025).
• Simulation and prediction: Deep-learning emulators (e.g., DeeptDCS) approximate current-distribution fields with <10% error vs finite-element methods, enabling real-time, individualized protocol design (Jia et al., 2022). Correlated CNN-based segmentation of subcortical targets further accelerates patient-specific modeling (Rashed et al., 2020).

Magnetic Stimulation

• Field–anatomo-functional mapping: FEM/BEM dosimetry, microsecond-scale circuit resolution (AMPS-TMS: 10 kHz bandwidth, <3% THD with 3 modules), and high-fidelity pulse shaping (arbitrary waveform synthesis; near-exponential voltage-resolution) provide tailored activation of axonal elements and plasticity pathways (Zhang et al., 8 Mar 2025, Gomez-Tames et al., 2020, Bai et al., 6 Jul 2025).
• Energetic and safety trade-offs: TMS enables sub-centimeter focality with 200–300× lower scalp-to-cortex field ratios (and less off-target skin stimulation) than tES for matched amplitude. However, this improvement requires mega-volt-ampere scale power and specialized electronics (Sheltraw et al., 2020).

Ultrasound-based NIBS

• Acoustic focusing and volumetric precision: OBUS achieves lateral FWHM of 152 μm, axial FWHM of 1.93 mm, preserving beam shape and intensity through the skull better than equivalent Gaussian sources, and activating neural circuits at up to 2.2 mm cortical depth in vivo (Li et al., 8 Jul 2025). OFUS further reduces focal spot size (83 μm) and energy density (0.6 mJ/cm²), four orders of magnitude below tFUS (Li et al., 2022).
• Transcranial efficiency: Bessel beams generated optoacoustically exhibit self-healing and aberration resistance—transmitting 18.7% of peak intensity through bone, versus 11% for conventional focused ultrasound, with improved preservation of spatial precision after skull traversal (Li et al., 8 Jul 2025).

Photobiomodulation

• Optical propagation and safety: 810 nm tPBM penetrates to ~1 cm depth, with less than 0.05 °C increase in cortical temperature during standard protocols (100 mW/cm² for 20 minutes), adhering to ΔT <1 °C safety limits (Ibrahimi et al., 2022).
• NIR-II advantages: 1064 nm (NIR-II) stimulation extends reach to deeper layers via reduced scattering, with no detectable EEG/MRI/cognitive side effects and a measurable decrease in serum neuron-specific enolase (NSE), indexing possible neuroprotection (Li et al., 13 Jul 2024).

3. Closed-loop, Individualized, and Robotic Platforms

• Closed-loop protocol optimization: Bayesian optimization frameworks—using real-time fMRI feedback and Gaussian process regression (RBF kernel, UCB acquisition function)—rapidly converge on maximally-effective tACS parameters (frequency, phase) for target network engagement, outperforming greedy approaches (expected improvement) (Lorenz et al., 2016).
• Robot-assisted TMS and navigation: Integrated systems (Robo-TMS) deliver TMS via 6–7 DOF manipulators with sub-2 mm/1.5° accuracy and closed-loop force control, supporting session-to-session reproducibility and safety. Automated calibration, marker-less registration, and deep-learning-based electric field models enable fast, personalized treatment planning (Bai et al., 6 Jul 2025, Gomez-Tames et al., 2020, Preiswerk et al., 2019).
• Multimodal mapping: Open-source neuronavigation systems (e.g., based on 3D Slicer) synchronize MRI, real-time head/applicator tracking, and stimulation modalities (TMS, tDCS, FUS), providing <1 mm spatial error across workflows (Preiswerk et al., 2019).

4. Mechanism of Action and Experimental Evidence

• Cellular and network effects: All NIBS modalities modulate excitability via polarization of neuronal membranes (ΔV_m ≈ −E · L · cos θ), influence spike timing/plasticity (tACS, tRNS, TMS pulse parameters), and can entrain or disrupt oscillatory dynamics. tPBM mechanisms center on cytochrome c oxidase activation, augmenting ATP production and modulating oxidative stress (Ibrahimi et al., 2022, Li et al., 13 Jul 2024).
• Empirical evidence for enhanced learning and therapeutic benefit: Anodal tDCS (1 mA, 10 min) over sensorimotor cortex reduces error and variability in surgical-skill acquisition (pattern cutting task), with increased task-related M1 activation detected via fNIRS. Effects persist for 1 month post-training, with tRNS and sham controls less effective (Gao et al., 2020).
• Primate and animal models: NHP studies show that TMS and tDCS perturb single-unit firing, synchronize with neural oscillations, and induce excitation-inhibition cycles in circuits such as the oculomotor network. Focused ultrasound modulates deep targets (e.g., superior colliculus) noninvasively, opening access to subcortical nodes (Lehmann et al., 2021, Li et al., 2022).

5. Spatial, Temporal, and Safety Constraints

• Spatial focality and off-target control: High-definition tDCS/tACS arrays and optimization (L1/L2 norm constraints and pointwise state constraints) reduce off-target electric field exposure by 4–6× relative to standard bipolar montages, enhancing alignment with anatomical normals and decreasing risk to non-target tissues (Wagner et al., 2015).
• Thermal and field-exposure limits: All modalities are constrained by maximum safe current and E-field density (≤0.05 mA/cm² for tES, MI<1.9 for ultrasound, ΔT<1 °C for tPBM), with empirical and simulation evidence confirming that modern protocols remain within established thresholds (Ibrahimi et al., 2022, Li et al., 13 Jul 2024, Li et al., 8 Jul 2025).
• Energetic and practical considerations: Electric stimulation is more energy-efficient at low frequencies but less focal; magnetic and acoustic modalities achieve high focality at the cost of increased device complexity and power demand (Sheltraw et al., 2020, Zhang et al., 8 Mar 2025).

6. Advances, Limitations, and Future Directions

• Technological innovation: Asymmetric modular TMS drive circuits offer near-exponential output-level granularity with fewer modules and lower distortion, enabling precise pulse shaping and selective neural activation (Zhang et al., 8 Mar 2025).
• Accessibility and reproducibility: Open-source toolchains for multimodal navigation, deep-learning emulation of field distributions, and end-to-end brain segmentation streamline translation from computational modeling to individualized therapy (Jia et al., 2022, Rashed et al., 2020, Preiswerk et al., 2019).
• Open questions: Critical unknowns include the optimal waveforms and montage parameters for deep, selective, and safe stimulation; the relationship between field properties and network plasticity/plasticity-driven outcomes; and the mechanisms mediating clinical benefits across diverse neurological disorders (Kish et al., 2 Apr 2024, Zibandepour et al., 2023).
• Toward closed-loop NIBS: Integration of real-time neuroimaging, Bayesian optimization, and adaptive parameter control promises fully individualized dosing and targeting, with subject-specific anatomical, functional, and behavioral endpoints guiding stimulation protocols (Lorenz et al., 2016, Bai et al., 6 Jul 2025).

Non-invasive brain stimulation thus represents a rapidly evolving, rigorously quantitative field, uniting theory, simulation, device engineering, and experimental neuroscience to enable increasingly precise, versatile, and safe modulation of brain function across research and clinical contexts.