Bayesian-enhanced closed-loop optimization of ultrasound protocols for targeted and precise neuromodulation
Andrea Boscutti, Valeria Grasso, Tommaso Di Ianni
2026 · 10.64898/2026.06.16.732762
Abstract
Low-intensity focused ultrasound (LIFU) is a promising neuromodulation modality, but challenges related to high response variability and the poorly understood parameter space undermine progress in clinical applications. To facilitate the development of therapeutic LIFU protocols, we developed an approach for Bayesian-enhanced adaptive control of ultrasound neuromodulation (BEACUN). BEACUN enables efficient, data-driven parameter mapping using a limited number of stimulation-response evaluations. We used functional ultrasound imaging (fUSI) to measure the neural responses to LIFU stimulation in real time, and we carried out in vivo experiments in rats to optimize and validate the performance of the BEACUN search. In live optimizations, we show that BEACUN produces more effective inhibitory LIFU neuromodulation protocols than conventional parameter exploration methods and converges to the optimal solution in 23 ± 3.67 stimulation-response evaluations. Our approach realizes a platform for efficient optimization of neuromodulation parameters that could pave the way for personalized LIFU protocol development in patients.
Abstract via europepmc.
Exposures
Exposure 1: superior colliculus (SC) fUSI-LIFU parameter mapping and live BEACUN optimization
Target: superior colliculus — “superior colliculus (SC)”
Device: other named manufacturer · Vermon, France · linear array, 15-MHz centre frequency, 128 elements, 0.1-mm pitch (driven by a Verasonics Vantage NXT 256-channel research scanner) ✓
| Waveform | pulsed | |
|---|---|---|
| Fundamental frequency (kHz) | 15,000 | ✓✓✓ |
| Pulse duration (ms) | not reported | ⚑ |
| Pulse repetition frequency (Hz) | 100, 100, 100, 10, 2swept | ✓✓✓ |
| Duty cycle (%) | 10, 5, 1, 1, 1swept | ✓✓✓ |
| Sonication duration (s) | 0.5 | ✓✓✓⚑ |
| Free-field pressure (kPa) | not reported | |
|---|---|---|
| Free-field Isppa (W/cm²) | not reported | |
| Free-field Ispta (W/cm²) | not reported | |
| In-situ estimate | deratingmean or range across subjects | |
| In-situ pressure (kPa) | 0, 2,200swept | ✓✓✓ |
| In-situ Isppa (W/cm²) | not reported | |
| In-situ Ispta (W/cm²) | not reported |
Repeated stimulation sessions tested 5 PRF-DC combinations (100 Hz/10%, 100 Hz/5%, 100 Hz/1%, 10 Hz/1%, 2 Hz/1%); pressure was varied in 6 steps between 0 and 2.2 MPa within each session. Each 110-s trial comprised a 20-s baseline, a 30-s period of interleaved fUSI/LIFU stimuli (during which a 500-ms LIFU pulse train alternated between two laterally-displaced foci, followed by a 750-ms delay before the next fUSI frame), and a 60-s post-stimulation period. Individual LIFU pulse duration was capped at 5 ms to prevent transducer overheating. In separate live BEACUN validation experiments (N=5 additional rats), pressure, PRF and DC were treated as continuous parameters and pressure was limited to 1.6 MPa.
Exposure 2: dorsal lateral geniculate (DLG) nucleus, live BEACUN optimization with concurrent visual stimulation
Target: lateral geniculate nucleus — “dorsal lateral geniculate (DLG) nucleus”
Device: other named manufacturer · Vermon, France · linear array, 15-MHz centre frequency, 128 elements, 0.1-mm pitch (driven by a Verasonics Vantage NXT 256-channel research scanner) ✓
| Waveform | pulsed | |
|---|---|---|
| Fundamental frequency (kHz) | 15,000 | ✓✓✓ |
| Pulse duration (ms) | not reportedimplied by duty cycle ÷ PRF: 1.0437 ms (not stated by the paper) | |
| Pulse repetition frequency (Hz) | 95.81 | ✓✓✓ |
| Duty cycle (%) | 10 | ✓✓✓ |
| Sonication duration (s) | 0.5 | ✓✓✓ |
| Free-field pressure (kPa) | not reported | |
|---|---|---|
| Free-field Isppa (W/cm²) | not reported | |
| Free-field Ispta (W/cm²) | not reported | |
| In-situ estimate | deratingmean or range across subjects | |
| In-situ pressure (kPa) | 1,000 | ✓✓✓⚑ |
| In-situ Isppa (W/cm²) | not reported | |
| In-situ Ispta (W/cm²) | not reported |
Live BEACUN optimization (N=5 rats) searched continuous PRF/DC/pressure parameters to suppress visual-evoked fUSI responses at the DLG target; pressure was limited to 1 MPa (max allowed). At the optimum found across sessions, PRF was 95.81 +/- 9.38 Hz, duty cycle was 10%, and pressure was 1 MPa (the maximum allowed).
Flags from extraction
n_subjects— Paper reports separate rat counts for the SC ground-truth parameter mapping (N=14, 48 sessions), the live SC BEACUN validation (N=5), and the live DLG BEACUN experiments (N=5); overlap between these cohorts is not stated, so counts are listed rather than summed.anaesthesia— Anaesthesia status is stated explicitly only for the DLG visual-stimulation sessions ('anesthetized rats were habituated in a dark chamber'); it is not stated for the SC fUSI-LIFU sessions.randomised— Randomisation refers to shuffling of stimulus/pressure/protocol order within and across sessions, not allocation to a control/treatment group (there is no control group in this methods/optimization paper).exposures[0].timing.pulse_duration_ms— Individual LIFU pulse duration is stated only as an upper limit (5 ms, to avoid transducer overheating), not as a value for each PRF/DC combination tested; per-condition pulse duration would require dividing duty cycle by PRF (arithmetic not performed).exposures[1].in_situ.pressure_kpa— The 1 MPa value for the DLG live BEACUN experiment is the maximum pressure allowed (a search-space bound) and also the pressure found at the optimum; it is not an independently reported free-standing exposure pressure.exposures[0].timing.sonication_duration_s— Sonication is delivered as two interleaved 500-ms LIFU pulse trains (alternating foci) with 750-ms gaps within the 30-s stimulation window; the 30 s value refers to this whole interleaved window, not one uninterrupted train.