First: Tianjin University's Haihe Brain-Computer Interface Lab Achieves High Spatiotemporal Resolution Decoding for Human Ultrasound Brain-Computer Interface
IT Home, November 22 news: The team from Tianjin University's Haihe Brain-Computer Interface Lab recently published its latest research findings in the internationally authoritative journal Science Advances.
The team proposed a highly robust ultrasound-modulated EEG signal decoding technique, achieving for the first time high-precision decoding of steady-state visual evoked potential (SSVEP) brain source signals at the level of human subject data.
This technique can reconstruct intracranial brain source waveforms with high fidelity and non-invasively under various complex and extreme scenarios, providing key support for the broad application of high spatiotemporal resolution ultrasound brain-computer interfaces.
According to the introduction, brain-computer interfaces, by establishing a direct information pathway between the brain and computers, are expected to realize the highest form of human-machine hybrid intelligence. However, traditional non-invasive brain-computer interfaces are limited by the low spatial resolution bottleneck of scalp EEG, which restricts the amount of brain information that can be read and severely constrains system performance. Ultrasound-modulated EEG technology is a new non-invasive EEG acquisition method based on the acoustoelectric effect, which combines both the high temporal resolution advantage of EEG and the high spatial resolution advantage of transcranial focused ultrasound. However, due to strong acoustic attenuation by individual skulls, ultrasound transcranial focal point shifts, and fluctuations in the activity intensity of deep neural sources, the acoustoelectric coupling effect is often in an extremely unstable state.
In practical applications of ultrasound brain-computer interfaces, due to problems such as the low signal-to-noise ratio of intracranial source signals collected extracranially, interference from multiple source signals, and non-uniform attenuation of transcranial ultrasound energy, the strength of acoustoelectric coupling signals fluctuates dramatically. This fluctuation causes the coupled signals to often be in complex modulation scenarios of weak acoustoelectric modulation (sideband energy at extremely weak noise levels) or excessive acoustoelectric modulation (sideband energy approaching or even exceeding the main peak energy). There is currently a lack of quantitative analysis metrics for these complex scenarios, as well as highly robust decoding algorithms that can non-invasively restore intracranial source signal waveforms under extreme distortion scenarios, seriously affecting the system performance and application effectiveness of ultrasound brain-computer interfaces.

To address this technical challenge, the team first proposed the Acoustoelectric Modulation Ratio (AMR) to define different acoustoelectric modulation scenarios, classifying modulation conditions into three scenarios—weak modulation, moderate modulation, and excessive modulation—based on acoustoelectric coupling depth, and further designed a Pulse Repetition Frequency (PRF) decoding algorithm applicable across scenarios.
Specifically, at the level of underlying mechanisms, the research broke through the previous scope of only qualitatively describing the acoustoelectric effect, using AMR as the core metric for defining acoustoelectric coupling depth. The AMR parameter not only intuitively reflects the coupling strength of the acoustic-electric components but also determines the performance boundary of ultrasound EEG signal decoding algorithms. Meanwhile, at the level of decoding algorithms, by extracting the instantaneous complex response at specific PRF frequencies, the research can effectively isolate low-frequency noise and overcome waveform distortion caused by different modulation depths, converting the spatial location labels provided by focused ultrasound into spatial information of brain source signals, thereby solving at the source the low spatial resolution bottleneck of traditional non-invasive scalp EEG caused by volume conduction effects.

The research first achieved sub-millimeter focal region control (HPBW as low as 0.79mm) based on a self-developed 128-element ultrasound phased array system, ensuring the precision of focused ultrasound spatial labeling. Then, in numerical simulations and brain-tissue-mimicking phantom experiments, it was demonstrated that the PRF decoding algorithm can achieve high-fidelity reconstruction of source signal waveforms in all three acoustoelectric modulation scenarios. Furthermore, based on the MetaBCI platform, a 10-participant dual-command SSVEP experiment was designed. Compared with traditional decoding algorithms such as envelope extraction, coherent demodulation, time-frequency analysis, and adaptive demodulation, under weak modulation (0 < AMR ≤ 0.6), moderate modulation (0.6 < AMR ≤ 1), and over-modulation (AMR > 1) states, the PRF algorithm achieved average decoding correlations of 0.84, 0.90, and 0.89, respectively, demonstrating excellent cross-scenario robustness.

https://www.science.org/doi/10.1126/sciadv.aei1954
Advertising statement: External redirect links contained in this article (including but not limited to hyperlinks, QR codes, passwords, and other forms) are used to convey more information and save selection time; results are for reference only. All articles on IT Home containing external links include this statement.
Related Articles
- 2026 Brain-Computer Interface Report: Global companies increase to 1,175, with China's three major city clusters gathering over 190
- BrainCo CEO Han Bicheng: In the future, people are expected to directly control phones and computers with their brains
- Chinese Academy of Sciences reveals that the brain can decode three-dimensional hand movement trajectories using only a small number of neurons, opening a new path for brain-computer interface R&D
- Huaqiangbei Brain-Computer Interface Industry Application Center recently opened, with over 60% of visitors being foreigners
- A new milestone: China's first non-invasive brain-computer interface project completes filing under Order No. 818
- Another important achievement: China-led international standard for brain-computer interface reference architecture officially released