A brain–computer interface (BCI) is a promising technology that can analyze brain signals and control a robot or computer according to a user’s intention. This paper introduces our studies to overcome the challenges of using BCIs in daily life. There are several methods to implement BCIs, such as sensorimotor rhythms (SMR), P300, and steady-state visually evoked potential (SSVEP). These methods have different pros and cons according to the BCI type. However, all these methods are limited in choice. Controlling the robot arm according to the intention enables BCI users can do various things. We introduced the study predicting three-dimensional arm movement using a non-invasive method. Moreover, the study was described compensating the prediction using an external camera for high accuracy. For daily use, BCI users should be able to turn on or off the BCI system because of the prediction error. The users should also be able to change the BCI mode to the efficient BCI type. The BCI mode can be transformed based on the user state. Our study was explained estimating a user state based on a brain’s functional connectivity and a convolutional neural network (CNN). Additionally, BCI users should be able to do various tasks, such as carrying an object, walking, or talking simultaneously. A multi-function BCI study was described to predict multiple intentions simultaneously through a single classification model. Finally, we suggest our view for the future direction of BCI study. Although there are still many limitations when using BCI in daily life, we hope that our studies will be a foundation for developing a practical BCI system.
Brain-computer interface (BCI) is a promising and very helpful technology. BCI studies have attempted to predict arm movements to control robotic arm depending on movement intentions. However, the low accuracy of movement prediction is a critical challenge in predicting arm movements. The aim of this study was to predict arm movement with high accuracy from non-invasive neural signals using the deep learning algorithm. We compared the prediction accuracies of the conventional method and long short-term memory (LSTM) using non-invasive MEG signals. This is the first study that applied LSTM to predict arm movements from non-invasive neural signals. The coefficients of correlation between real signals and signals predicted by the MLR on the x-, y-, and z-axes were 0.677 ± 0.139 (mean ± SD), 0.689 ± 0.140, and 0.785 ± 0.103, respectively. The coefficients of correlation between real signals and signals predicted by the LSTM on the x-, y-, and z-axes were 0.978 ± 0.004, 0.980 ± 0.005, and 0.980 ± 0.008, respectively. The prediction accuracy was highly improved using the LSTM algorithm. Our results suggest that highly accurate prediction of arm movement is possible without surgery using the deep learning algorithm. We expect that the deep learning algorithm will facilitate the control of a robot arm using non-invasive signals in real life.
Studying the motor-control mechanisms of the brain is critical in academia and also has practical implications because techniques such as brain-computer interfaces (BCIs) can be developed based on brain mechanisms. Magnetoencephalography (MEG) signals have the highest spatial resolution (~3 mm) and temporal resolution (~1 ms) among the non-invasive methods. Therefore, the MEG is an excellent modality for investigating brain mechanisms. However, publicly available MEG data remains scarce due to expensive MEG equipment, requiring a magnetically shielded room, and high maintenance costs for the helium gas supply. In this study, we share the 306-channel MEG and 3-axis accelerometer signals acquired during three-dimensional reaching movements. Additionally, we provide analysis results and MATLAB codes for time-frequency analysis, F-value time-frequency analysis, and topography analysis. These shared MEG datasets offer valuable resources for investigating brain activities or evaluating the accuracy of prediction algorithms. To the best of our knowledge, this data is the only publicly available MEG data measured during reaching movements.
Decoding neural signals into control outputs has been a key to the development of brain-computer interfaces (BCIs). While many studies have identified neural correlates of kinematics or applied advanced machine learning algorithms to improve decoding performance, relatively less attention has been paid to optimal design of decoding models. For generating continuous movements from neural activity, design of decoding models should address how to incorporate movement dynamics into models and how to select a model given specific BCI objectives. Considering nonlinear and independent speed characteristics, we propose a hybrid Kalman filter to decode the hand direction and speed independently. We also investigate changes in performance of different decoding models (the linear and Kalman filters) when they predict reaching movements only or predict both reach and rest. Our offline study on human magnetoencephalography (MEG) during point-to-point arm movements shows that the performance of the linear filter or the Kalman filter is affected by including resting states for training and predicting movements. However, the hybrid Kalman filter consistently outperforms others regardless of movement states. The results demonstrate that better design of decoding models is achieved by incorporating movement dynamics into modeling or selecting a model according to decoding objectives.
The brain–computer interface (BCI) is a promising technology where a user controls a robot or computer by thinking with no movement. There are several underlying principles to implement BCI, such as sensorimotor rhythms, P300, steady-state visually evoked potentials, and directional tuning. Generally, different principles are applied to BCI depending on the application, because strengths and weaknesses vary according to each BCI method. Therefore, BCI should be able to predict a user state to apply suitable principles to the system. This study measured electroencephalography signals in four states (resting, speech imagery, leg-motor imagery, and hand-motor imagery) from 10 healthy subjects. Mutual information from 64 channels was calculated as brain connectivity. We used a convolutional neural network to predict a user state, where brain connectivity was the network input. We applied five-fold cross-validation to evaluate the proposed method. Mean accuracy for user state classification was 88.25 ± 2.34%. This implies that the system can change the BCI principle using brain connectivity. Thus, a BCI user can control various applications according to their intentions.
Transformer models have revolutionized fields like Natural Language Processing (NLP) by enabling machines to accurately understand and generate human language. However, these models’ inherent complexity and limited interpretability pose barriers to their broader adoption. To address these challenges, we propose a simplified query-only attention mechanism specifically for encoder-based transformer models to reduce complexity and improve interpretability. Unlike conventional attention mechanisms, which rely on query (Q), key (K), and value (V) vectors, our method uses only the Q vector for attention calculation. This approach reduces computational complexity while maintaining the model’s ability to capture essential relationships, enhancing interpretability. We evaluated the proposed query-only attention on an EEG conformer model, a state-of-the-art architecture for EEG signal classification. We demonstrated that it performs comparably to the original QKV attention mechanism, while simplifying the model’s architecture. Our findings suggest that query-only attention offers a promising direction for the development of more efficient and interpretable transformer-based models, with potential applications across various domains beyond NLP.