← Explore

Ultrasound system for precise neuromodulation of human deep brain circuits

Eleanor Martin, Morgan Roberts, Ioana F. Grigoras, Olivia Wright, Tulika Nandi, Sebastian W. Rieger, Jon Campbell, Tim den Boer, Ben T. Cox, Charlotte J. Stagg, Bradley E. Treeby

Nature Communications 2025, 16 · 10.1038/s41467-025-63020-1

human healthyhealthyfmri

Abstract

We introduce an advanced transcranial ultrasound stimulation (TUS) system for precise deep brain neuromodulation, featuring a 256-element helmet-shaped transducer array (555 kHz), stereotactic positioning, individualised planning, and real-time fMRI monitoring. Experiments demonstrated selective modulation of the lateral geniculate nucleus (LGN) and connected visual cortex regions. Participants showed significantly increased visual cortex activity during concurrent TUS and visual stimulation, with high cross-individual reproducibility. A theta-burst TUS protocol produced robust neuromodulatory effects, decreasing visual cortex activity for at least 40 min post-stimulation. Control experiments confirmed these effects were specific to the targeted LGN. Our findings reveal this system's potential to non-invasively modulate deep brain circuits with unprecedented precision and specificity, offering new avenues for studying brain function and developing targeted therapies for neurological and psychiatric disorders, with transformative potential for both research and clinical applications.

Abstract via europepmc.

Specieshuman
Subjects7 participants
Sessions per subject4
Randomisedyes
Blindingsingle
Sham / controlinactive transducer, active control site
Auditory controlramped pulses
Readout timingboth
Anaesthesianot applicable
Readoutsfmri
Direction of effectbidirectionalOnline TUS to the LGN (300 ms pulses every 3 s, 775 kPa) increased visually-evoked activity in ipsilateral V1 during stimulation; offline theta-burst TUS to the LGN (20 ms pulses at 5 Hz for 80 s) instead decreased visually-evoked V1 activity ~40 min after stimulation, with the decrease no longer significant ~140 min after stimulation.
Adverse eventsnone observedAfter each session, participants were asked to report any changes in visual perception; no changes were reported by any participant. Simulations predicted a maximum brain temperature rise of less than 0.2 degC and MI_TC below 1.9, within ITRUSST safety limits.

Exposures

Exposure 1: TUS to left LGN (online block-design and offline theta-burst experiments)

Target: lateral geniculate nucleus — “lateral geniculate nucleus (LGN)
Device: Sonic Concepts · Sonic Concepts (custom transducer elements) · Custom 256-element helmet array

Pulse timing
Waveformpulsed, theta_burst
Fundamental frequency (kHz)555✓✓
Pulse duration (ms)300, 20swept✓✓
Pulse repetition frequency (Hz)0.33, 5swept?
Duty cycle (%)10✓✓
Sonication duration (s)15, 80swept✓✓
Pressure and intensity, by domain
Free-field pressure (kPa)2,580✓✓
Free-field Isppa (W/cm²)not reported
Free-field Ispta (W/cm²)not reported
In-situ estimatesimulationmean or range across subjects
In-situ pressure (kPa)775✓✓
In-situ Isppa (W/cm²)20✓✓
In-situ Ispta (W/cm²)not reported
Protocol, in the paper’s words

Online experiment: active TUS delivered as 300 ms pulses every 3 s (10 ms cosine ramp up/down) during 10 of 20 checkerboard blocks (15 s stimulus blocks, 9 s rest), interleaved with sham (system powered on, no ultrasound); up to six on-target runs per day over two visits, plus off-target (MDN) control runs in three participants. Offline experiment: theta-burst protocol of 20 ms ultrasound pulses delivered at 5 Hz (every 200 ms; duty cycle 10%; 1 ms ramp up/down) for a total train duration of 80 s, delivered once to the LGN and once to the active-control MDN site in four participants, with fMRI acquired at baseline, ~20-40 min (early) and ~140-160 min (late) after stimulation.

Flags from extraction

  • exposures[0].timing.pulse_duration_msReported as one exposure (single target/frequency) per instructions; list combines the online 300 ms pulse with the offline 20 ms theta-burst pulse, two structurally different protocols.
  • exposures[0].timing.pulse_repetition_frequency_hzOnline value (0.33 Hz) is converted from the stated '300 ms pulses every 3 s' interval (a period-to-frequency conversion), not stated directly as a PRF by the authors; offline PRF (5 Hz) is stated directly.
  • exposures[0].timing.duty_cycle_pct10% duty cycle is stated only for the offline theta-burst protocol; the online single-pulse-per-block protocol has no stated duty cycle, so only one value is given despite two protocols.
  • exposures[0].timing.sonication_duration_sOnline value (15 s) is the duration of one active checkerboard/TUS block (Fig. S7 'Pulse Train'); offline value (80 s) is the total theta-burst train duration; these are not the same kind of quantity.
  • exposures[0].free_field.pressure_kpaFree-field spatial peak pressure amplitude (2.58 +/- 0.06 MPa) taken from the Left LGN row of Supplementary Fig. S7 table; recorded as the mean.
  • exposures[0].in_situ.pressure_kpaIn-situ pressure (775 +/- 163 kPa) is the achieved value across participants/sessions from the Supplementary Fig. S7 table, coinciding with the 775 kPa target set point stated in the main text.
  • blindingOnline experiment is described as single-blind; the offline (theta-burst) experiment is explicitly described as unblinded. The single value recorded reflects the primary/online experiment only.
  • sham_typeIncludes both the online experiment's device-powered-but-inactive sham and the active-control-site (MDN) condition used in both experiments.