A high-performance speech neuroprosthesis

Nature, Journal Year: 2023, Volume and Issue: 620(7976), P. 1031 - 1036

Published: Aug. 23, 2023

Speech brain-computer interfaces (BCIs) have the potential to restore rapid communication people with paralysis by decoding neural activity evoked attempted speech into text

Language: Английский

---

The speech neuroprosthesis

Nature reviews. Neuroscience, Journal Year: 2024, Volume and Issue: 25(7), P. 473 - 492

Published: May 14, 2024

Language: Английский

---

A high-performance brain–computer interface for finger decoding and quadcopter game control in an individual with paralysis

Nature Medicine, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 20, 2025

Abstract People with paralysis express unmet needs for peer support, leisure activities and sporting activities. Many within the general population rely on social media massively multiplayer video games to address these needs. We developed a high-performance, finger-based brain–computer-interface system allowing continuous control of three independent finger groups, which thumb can be controlled in two dimensions, yielding total four degrees freedom. The was tested human research participant tetraplegia due spinal cord injury over sequential trials requiring fingers reach hold targets, an average acquisition rate 76 targets per minute completion time 1.58 ± 0.06 seconds—comparing favorably prior animal studies despite twofold increase decoded More importantly, positions were then used virtual quadcopter—the number-one restorative priority participant—using brain-to-finger-to-computer interface allow dexterous navigation around fixed- random-ringed obstacle courses. expressed or demonstrated sense enablement, recreation connectedness that addresses many people paralysis.

Language: Английский

---

A high-performance speech neuroprosthesis

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 21, 2023

Abstract Speech brain-computer interfaces (BCIs) have the potential to restore rapid communication people with paralysis by decoding neural activity evoked attempted speaking movements into text 1,2 or sound 3,4 . Early demonstrations, while promising, not yet achieved accuracies high enough for of unconstrainted sentences from a large vocabulary 1–7 Here, we demonstrate first speech-to-text BCI that records spiking intracortical microelectrode arrays. Enabled these high-resolution recordings, our study participant, who can no longer speak intelligibly due amyotrophic lateral sclerosis (ALS), 9.1% word error rate on 50 (2.7 times fewer errors than prior state art speech 2 ) and 23.8% 125,000 (the successful demonstration large-vocabulary decoding). Our decoded at 62 words per minute, which is 3.4 faster record any kind 8 begins approach speed natural conversation (160 minute 9 ). Finally, highlight two aspects code are encouraging BCIs: spatially intermixed tuning articulators makes accurate possible only small region cortex, detailed articulatory representation phonemes persists years after paralysis. These results show feasible path forward using BCIs speak.

Language: Английский

---

Using adversarial networks to extend brain computer interface decoding accuracy over time

eLife, Journal Year: 2023, Volume and Issue: 12

Published: Aug. 23, 2023

Existing intracortical brain computer interfaces (iBCIs) transform neural activity into control signals capable of restoring movement to persons with paralysis. However, the accuracy 'decoder' at heart iBCI typically degrades over time due turnover recorded neurons. To compensate, decoders can be recalibrated, but this requires user spend extra and effort provide necessary data, then learn new dynamics. As neurons change, one think underlying intent signal being expressed in changing coordinates. If a mapping computed between different coordinate systems, it may possible stabilize original decoder's from behavior without recalibration. We previously proposed method based on Generalized Adversarial Networks (GANs), called 'Adversarial Domain Adaptation Network' (ADAN), which aligns distributions latent within low-dimensional manifolds. we tested ADAN only very limited dataset. Here propose Cycle-Consistent (Cycle-GAN), full-dimensional recordings. both Cycle-GAN data multiple monkeys behaviors compared them third, quite Procrustes alignment axes provided by Factor Analysis. All three methods are unsupervised require little making practical real life. Overall, had best performance was easier train more robust than ADAN, ideal for stabilizing systems time.

Language: Английский

---

Decoding motor plans using a closed-loop ultrasonic brain–machine interface

Nature Neuroscience, Journal Year: 2023, Volume and Issue: 27(1), P. 196 - 207

Published: Nov. 30, 2023

Brain-machine interfaces (BMIs) enable people living with chronic paralysis to control computers, robots and more nothing but thought. Existing BMIs have trade-offs across invasiveness, performance, spatial coverage spatiotemporal resolution. Functional ultrasound (fUS) neuroimaging is an emerging technology that balances these attributes may complement existing BMI recording technologies. In this study, we use fUS demonstrate a successful implementation of closed-loop ultrasonic BMI. We streamed data from the posterior parietal cortex two rhesus macaque monkeys while they performed eye hand movements. After training, controlled up eight movement directions using also developed method for pretraining previous sessions. This enabled immediate on subsequent days, even those occurred months apart, without requiring extensive recalibration. These findings establish feasibility BMIs, paving way new class less-invasive (epidural) generalize extended time periods promise restore function neurological impairments.

Language: Английский

---

A real-time, high-performance brain-computer interface for finger decoding and quadcopter control

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Feb. 8, 2024

People with paralysis express unmet needs for peer support, leisure activities, and sporting activities. Many within the general population rely on social media massively multiplayer video games to address these needs. We developed a high-performance finger brain-computer-interface system allowing continuous control of 3 independent groups 2D thumb movements. The was tested in human research participant over sequential trials requiring fingers reach hold targets, an average acquisition rate 76 targets/minute completion time 1.58 ± 0.06 seconds. Performance compared favorably previous animal studies, despite 2-fold increase decoded degrees-of-freedom (DOF). Finger positions were then used 4-DOF velocity virtual quadcopter, demonstrating functionality both fixed random obstacle courses. This approach shows promise controlling multiple-DOF end-effectors, such as robotic or digital interfaces work, entertainment, socialization.

Language: Английский

---

Assistive sensory-motor perturbations influence learned neural representations

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: March 20, 2024

Task errors are used to learn and refine motor skills. We investigated how task assistance influences learned neural representations using Brain-Computer Interfaces (BCIs), which map activity into movement via a decoder. analyzed cortex as monkeys practiced BCI with decoder that adapted improve or maintain performance over days. The dimensionality of the population neurons controlling remained constant increased learning, counter expected trends from learning. Yet, time, information was contained in smaller subset modes. Moreover, ultimately stored modes occupied small fraction variance. An artificial network model suggests adaptive decoders contribute forming these compact representations. Our findings show assistive manipulate error for long-term learning computations, like credit assignment, informs our understanding has implications designing real-world BCIs.

Language: Английский

---

Long-term unsupervised recalibration of cursor BCIs

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Feb. 4, 2023

Abstract Intracortical brain-computer interfaces (iBCIs) require frequent recalibration to maintain robust performance due changes in neural activity that accumulate over time. Compensating for this nonstationarity would enable consistently high without the need supervised periods, where users cannot engage free use of their device. Here we introduce a hidden Markov model (HMM) infer what targets are moving toward during iBCI use. We then retrain system using these inferred targets, enabling unsupervised adaptation changing activity. Our approach outperforms state art large-scale, closed-loop simulations two months and with human user one month. Leveraging an offline dataset spanning five years recordings, further show how recently proposed data distribution-matching approaches fail long time scales; only target-inference methods appear capable long-term recalibration. results demonstrate task structure can be used bootstrap noisy decoder into highly-performant one, thereby overcoming major barriers clinically translating BCIs.

Language: Английский

---

Exploring Synergies in Brain-Machine Interfaces: Compression vs. Performance

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 3, 2025

Abstract Individuals with severe neurological injuries often rely on assistive technologies, but current methods have limitations in accurately decoding multi-degree-of-freedom (DoF) movements. Intracortical brain-machine interfaces (iBMIs) use neural signals to provide a more natural control method, currently struggle higher-DoF movements—something the brain handles effortlessly. It has been theorized that simplifies high-DoF movement through muscle synergies, which link multiple muscles function as single unit. These synergies studied using dimensionality reduction techniques like principal component analysis (PCA), non-negative matrix factorization (NMF), and demixed PCA (dPCA) successfully used reduce noise improve offline decoder stability non-invasive applications. However, their effectiveness improving generalizability for implanted recordings across varied tasks is unclear. Here, we evaluated if can enhance iBMI performance non-human primates performing two-DoF finger task. Specifically, tested PCA, dPCA, NMF could compress denoise data generalization tasks. Our results showed while all effectively compressed minimal loss accuracy, none improved denoising. Additionally, of enhanced findings suggest aid compression, alone it may not reveal “true” space needed or generalizability. Further research required determine whether are optimal framework alternative approaches robustness Significance Statement Many researchers believe represent fundamental strategy interface (BMI) performance. extracted techniques, thought simplify complex data, efficiency accuracy BMI systems. In our study, dexterous We found these high-dimensional they did denoising generalize well different contexts. Instead, highest was achieved when available suggesting although useful adaptability

Language: Английский

---