Wrist Signals Decoded: A Brain-Controlled Leap for Neuromorphic Models

In a fascinating study that just came out, the ability to control robotic arms or prosthetic hands with thought alone is being further propelled toward reality through advancements in Brain-Computer Interfaces (BCIs). This research, published in Biomimetics, has significantly improved the accuracy of BCIs by focusing on the neural signals associated with wrist movements, offering valuable insights that can also inform the development of neuromorphic computing systems.
Researchers Abdul-Khaaliq Mohamed and Vered Aharonson looked into how electroencephalography (EEG), a non-invasive technique that measures brain electrical activity, can be used to distinguish between wrist extension (WE) and wrist flexion (WF) movements. Their goal was to improve upon the ability of BCIs to control prosthetic or orthotic hands, helping individuals with motor impairments regain independence. Simultaneously, this research contributes to a deeper understanding of neural motor control, which is essential for building more biologically realistic neuromorphic systems.
Decoding Neural Signals
The study involved 14 participants who performed repeated isometric (muscle contraction without movement) wrist extension and flexion exercises. In ensuring consistency, the force, speed, and range of motion were carefully regulated using a purpose designed dynamometer (IsoReg). The researchers used a 128-channel EEG system to capture the brain activity during these movements.
The EEG data were then analyzed in several steps. First, the data were divided into four frequency bands: delta/theta (1-7 Hz), mu/beta (7-35 Hz), low gamma (35-49 Hz), and high gamma (51-90 Hz). These bands represent different types of brain activity. Next, independent component analysis (ICA), a technique to separate mixed signals into independent sources, was used to isolate signals originating from seven specific cortical regions: bilateral (both left and right) hand homunculus of the primary motor cortex (M1-H), bilateral prefrontal cortex (PFC), bilateral ventral premotor cortex (PMv), and the supplementary motor area (SMA).
To extract meaningful information from these signals, the researchers used a time-frequency algorithm. This algorithm breaks down the signals into their constituent frequencies over time, revealing patterns associated with the wrist movements. Features were selected using the Bhattacharyya distance (BD), a measure of separability between two classes. Finally, they used Mahalanobis distance (MD) clustering, a statistical technique that considers data correlations, to classify the extracted features and determine whether they corresponded to wrist extension or flexion.
Key Findings
The study revealed several significant findings. The researchers successfully classified bilateral (both wrists moving together) wrist extension and flexion movements with an impressive accuracy of 90.68%. Unilateral (one wrist moving) movements were classified with an accuracy of 69.80%. All four frequency bands contained information relevant to motor control, suggesting that BCIs and potentially neuromorphic systems can benefit from analyzing a wide range of brain activity. All seven cortical regions of interest contributed to the classification, indicating that motor control involves a distributed network of brain areas. Bilateral movements relied more on the mu and beta frequency bands, while unilateral movements favored the gamma frequency bands. This suggests that different brain mechanisms are involved in these two types of movements. The prefrontal cortex (PFC) showed the most frequent activity, suggesting its role in monitoring and coordinating the regulated isometric movements.
Implications for BCIs and Neuromorphic Computing
These findings have important implications for the development of more accurate and effective BCIs. By demonstrating the ability to decode wrist movements with high accuracy, the study paves the way for improved control of prosthetic and orthotic devices.
The results also highlight the significance of analyzing multiple frequency bands and cortical regions, which is relevant for both BCI and neuromorphic systems. Understanding the distributed nature of motor control and the role of different frequency bands can inform the design of more biologically realistic neuromorphic models. The signal processing techniques used in this study, such as time-frequency analysis and ICA, could be adapted for implementation in neuromorphic hardware and software.
Continuing Research and Future Directions
This research represents a significant leap forward in the field of BCI and contributes to our understanding of neural motor control, which is crucial for neuromorphic computing. Future research will explore the use of these techniques in real-world BCI applications, such as controlling robotic arms or prosthetic hands in everyday tasks. The integration of audio and other sensory input could also be explored to make BCI systems even more robust. Future research will also explore dynamic causal modeling (DCM) to further understand the causal cross frequency communication between cortical regions, providing more data for neuromorphic models.
By continuing to unravel the mysteries of brain activity, researchers are bringing us closer to a future where individuals with motor impairments can regain their independence and where neuromorphic systems can more accurately replicate brain function.
Citation:
Mohamed, A.-K., & Aharonson, V. (2025). Single-Trial Electroencephalography Discrimination of Real, Regulated, Isometric Wrist Extension and Wrist Flexion. Biomimetics, 10(3), 187. https://doi.org/10.3390/biomimetics10030187
NeuromorphicCore

