Closed-loop transcranial ultrasound stimulation based on NREM and REM sleep for bidirectional modulation of sleep neural oscillation and memory
Teng Wang, Mengran Wang, Yi Yuan
Journal of Neural Engineering 2026 · 10.1088/1741-2552/ae584a
Abstract
Objective. sleep plays an important role in memory integration. Closed-loop physical stimulation during rapid eye movement (REM) or non-REM (NREM) sleep can modulate neural oscillations and associated memory functions. However, the impact of closed-loop transcranial ultrasound stimulation (CLTUS), a non-invasive technique with high spatial resolution and deep penetration, on sleep-specific oscillations and memory remains unclear. Approach. In this study, we developed a CLTUS system using deep learning to target NREM and REM sleep oscillations and evaluated sleep neural oscillations and memory ability. Main results. Our findings revealed that CLTUS of the slow oscillation up-state during NREM sleep reduced NREM-specific neural activity and significantly weakened spatial and fear memory abilities. These changes were positively correlated with the induced neural dynamics. Conversely, during REM sleep, CLTUS delivered at the theta peak enhanced REM-specific neural activity in both healthy and Alzheimer's disease model mice. Consequently, spatial memory was significantly improved in both groups, and this improvement was closely related to ultrasound-induced neural activity. Significance. CLTUS based on NREM and REM sleep exerts a bidirectional modulation effect on sleep neural oscillations and memory ability, providing important guidance for selecting and optimizing stimulation protocols for modulating sleep and memory.
Abstract via europepmc.
Exposures
Exposure 1: Closed-loop TUS, motor cortex (NREM SO up-state; also pre-sleep)
Target: primary motor cortex — “motor cortex”
Device: Olympus / Panametrics · Olympus · V301-SU (unfocused; system TUNSS-1) ✓
| Waveform | pulsed | |
|---|---|---|
| Fundamental frequency (kHz) | 500 | ✓✓✓ |
| Pulse duration (ms) | 50 | ✓✓✓⚑ |
| Pulse repetition frequency (Hz) | 1 | ✓✓✓ |
| Duty cycle (%) | 5pulse duration × PRF gives 5% | ✓✓✓ |
| Sonication duration (s) | 900, 3,600swept | ✓✓✓⚑ |
| Free-field pressure (kPa) | not reported | |
|---|---|---|
| Free-field Isppa (W/cm²) | not reported | |
| Free-field Ispta (W/cm²) | not reported | |
| In-situ estimate | not reported | |
| In-situ pressure (kPa) | not reported | |
| In-situ Isppa (W/cm²) | not reported | |
| In-situ Ispta (W/cm²) | not reported | |
| Pressure, domain unspecified (kPa) | 310 | ✓✓✓⚑ |
| Isppa, domain unspecified (W/cm²) | 3.2 | ✓✓✓ |
| Ispta, domain unspecified (W/cm²) | 0.16 | ✓✓✓ |
Closed-loop: a 50 ms pulse triggered at each detected SO up-state (NREM) or theta peak (REM, 1 Hz scan), for 1 h from the first stimulus within a 2 h window, daily for seven consecutive days. Pre-sleep comparison group (motor cortex) received 15 min daily before an 8 h sleep recording.
Exposure 2: Closed-loop TUS, hippocampus (REM theta peak)
Target: hippocampus — “hippocampus”
Device: Olympus / Panametrics · Olympus · V301-SU (unfocused; system TUNSS-1) ✓
| Waveform | pulsed | |
|---|---|---|
| Fundamental frequency (kHz) | 500 | ✓✓✓ |
| Pulse duration (ms) | 50 | ✓✓✓ |
| Pulse repetition frequency (Hz) | 1 | ✓✓✓ |
| Duty cycle (%) | 5pulse duration × PRF gives 5% | ✓✓✓ |
| Sonication duration (s) | 3,600 | ✓✓✓ |
| Free-field pressure (kPa) | not reported | |
|---|---|---|
| Free-field Isppa (W/cm²) | not reported | |
| Free-field Ispta (W/cm²) | not reported | |
| In-situ estimate | not reported | |
| In-situ pressure (kPa) | not reported | |
| In-situ Isppa (W/cm²) | not reported | |
| In-situ Ispta (W/cm²) | not reported | |
| Pressure, domain unspecified (kPa) | 310 | ✓✓✓ |
| Isppa, domain unspecified (W/cm²) | 3.2 | ✓✓✓ |
| Ispta, domain unspecified (W/cm²) | 0.16 | ✓✓✓ |
Closed-loop: a 50 ms pulse triggered at each detected SO up-state (NREM) or theta peak (REM, 1 Hz scan), for 1 h from the first stimulus within a 2 h window, daily for seven consecutive days. Pre-sleep comparison group (motor cortex) received 15 min daily before an 8 h sleep recording.
Flags from extraction
n_subjects— 60 is the sum of the two enrolled strains (48 + 12), including sham and pre-sleep animals; number actually exposed to ultrasound not stated (figure legends give n = 6 per group).exposures[0].unspecified_domain.pressure_kpa— Hydrophone-measured values; the paper does not say whether water or through skull, so placed in unspecified_domain. Same values given for both targets.exposures[0].timing.sonication_duration_s— Stimulation is a triggered 50 ms pulse at each event; 3600 s is the 1 h delivery period from the first stimulus, not a continuous train.exposures[0].timing.pulse_duration_ms— Text says 'stimulation duration of 50 ms' and elsewhere 'pulse duration (50 ms)'; taken as pulse duration.exposures[0].label— One exposure per target; motor cortex entry also covers the pre-sleep stimulation group with the same parameters.