Closed-Loop Transcranial Ultrasound Stimulation Based on Deep Learning Effectively Suppresses Epileptic Seizures in Mice
Na Pang, Jingyan Sun, Hailin Zhang, Rong Chen, Jiaqing Yan, Yi Yuan
IEEE Transactions on Neural Systems and Rehabilitation Engineering 2025 · 10.1109/tnsre.2025.3589089
Abstract
Transcranial ultrasound stimulation is a non-invasive neuromodulation technique characterized by its high spatial resolution and penetration depth, and it has shown an inhibitory effect on epilepsy. However, current applications predominantly employ open-loop transcranial ultrasound stimulation, which lacks the capacity to dynamically respond to seizures. In the present study, we designed and implemented a closed-loop transcranial ultrasound stimulation (CTUS) system comprising a signal acquisition module, a signal preprocessing module, a deep learning network model-based epileptic signal recognition module, and an ultrasound stimulation module to enable real-time detection and ultrasound intervention in the hippocampus of penicillin-induced epileptic mice. The results indicated that the CTUS system could accurately identify epileptic signals, significantly reduce the seizure firing rate, decrease the power intensity and phase-amplitude coupling, and enhance the sample entropy. These findings demonstrated that the deep learning-based CTUS system was efficient in suppressing seizures in mice.
Abstract via europepmc.
Exposures
Exposure 1: Closed-loop TUS (CTUS) of hippocampus
Target: hippocampus — “hippocampus of penicillin-induced epileptic mice”
Device: Olympus / Panametrics · Olympus · V302-SU ✓
| Waveform | pulsed | |
|---|---|---|
| Fundamental frequency (kHz) | 1,000 | ✓✓✓ |
| Pulse duration (ms) | not reportedimplied by duty cycle ÷ PRF: 1 ms (not stated by the paper) | |
| Pulse repetition frequency (Hz) | 500 | ✓✓✓ |
| Duty cycle (%) | 50 | ✓✓✓ |
| Sonication duration (s) | 5 | ✓✓✓⚑ |
| 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) | 330 | ✓✓✓⚑ |
| Isppa, domain unspecified (W/cm²) | 3.6 | ✓✓✓⚑ |
Closed loop: LFP segments identified by a deep learning model at 0.2 s intervals; on detecting a seizure signal a 5 s TUS (1 MHz, PRF 500 Hz, DC 50%, 0.33 MPa) was triggered, then recognition resumed and the cycle repeated. Open-loop comparison group received continuous 15 min TUS with the same parameters.
Flags from extraction
exposures[0].unspecified_domain.isppa_w_cm2— Isppa 3.6 W/cm2 is described as calculated; domain not stated.exposures[0].unspecified_domain.pressure_kpa— Acoustic pressure 0.33 MPa, domain not stated.n_subjects— Ten mice allocated to CTUS and Sham (five each); the open-loop OTUS group (n = 5) is compared with a previous study. Only the CTUS group is counted as sonicated.exposures[0].timing.sonication_duration_s— Each triggered stimulation lasted 5 s; number and spacing depend on detected seizures.