Brain-machine interface based on deep learning to control asynchronously a lower-limb robotic exoskeleton: a case-of-study.

BACKGROUND: This research focused on the development of a motor imagery (MI) based brain-machine interface (BMI) using deep learning algorithms to control a lower-limb robotic exoskeleton. The study aimed to overcome the limitations of traditional BMI approaches by leveraging the advantages of deep learning, such as automated feature extraction and transfer learning. The experimental protocol to evaluate the BMI was designed as asynchronous, allowing subjects to perform mental tasks at their own will. METHODS: A total of five healthy able-bodied subjects were enrolled in this study to participate in a series of experimental sessions. The brain signals from two of these sessions were used to develop a generic deep learning model through transfer learning. Subsequently, this model was fine-tuned during the remaining sessions and subjected to evaluation. Three distinct deep learning approaches were compared: one that did not undergo fine-tuning, another that fine-tuned all layers of the model, and a third one that fine-tuned only the last three layers. The evaluation phase involved the exclusive closed-loop control of the exoskeleton device by the participants' neural activity using the second deep learning approach for the decoding. RESULTS: The three deep learning approaches were assessed in comparison to an approach based on spatial features that was trained for each subject and experimental session, demonstrating their superior performance. Interestingly, the deep learning approach without fine-tuning achieved comparable performance to the features-based approach, indicating that a generic model trained on data from different individuals and previous sessions can yield similar efficacy. Among the three deep learning approaches compared, fine-tuning all layer weights demonstrated the highest performance. CONCLUSION: This research represents an initial stride toward future calibration-free methods. Despite the efforts to diminish calibration time by leveraging data from other subjects, complete elimination proved unattainable. The study's discoveries hold notable significance for advancing calibration-free approaches, offering the promise of minimizing the need for training trials. Furthermore, the experimental evaluation protocol employed in this study aimed to replicate real-life scenarios, granting participants a higher degree of autonomy in decision-making regarding actions such as walking or stopping gait.

Exploring EEG-based motor imagery decoding: a dual approach using spatial features and spectro-spatial Deep Learning model IFNet.

Introduction: In recent years, the decoding of motor imagery (MI) from electroencephalography (EEG) signals has become a focus of research for brain-machine interfaces (BMIs) and neurorehabilitation. However, EEG signals present challenges due to their non-stationarity and the substantial presence of noise commonly found in recordings, making it difficult to design highly effective decoding algorithms. These algorithms are vital for controlling devices in neurorehabilitation tasks, as they activate the patient's motor cortex and contribute to their recovery. Methods: This study proposes a novel approach for decoding MI during pedalling tasks using EEG signals. A widespread approach is based on feature extraction using Common Spatial Patterns (CSP) followed by a linear discriminant analysis (LDA) as a classifier. The first approach covered in this work aims to investigate the efficacy of a task-discriminative feature extraction method based on CSP filter and LDA classifier. Additionally, the second alternative hypothesis explores the potential of a spectro-spatial Convolutional Neural Network (CNN) to further enhance the performance of the first approach. The proposed CNN architecture combines a preprocessing pipeline based on filter banks in the frequency domain with a convolutional neural network for spectro-temporal and spectro-spatial feature extraction. Results and discussion: To evaluate the approaches and their advantages and disadvantages, EEG data has been recorded from several able-bodied users while pedalling in a cycle ergometer in order to train motor imagery decoding models. The results show levels of accuracy up to 80% in some cases. The CNN approach shows greater accuracy despite higher instability.

Design and Evaluation of a Potential Non-Invasive Neurostimulation Strategy for Treating Persistent Anosmia in Post-COVID-19 Patients.

A new pandemic was declared at the end of 2019 because of coronavirus disease 2019 (COVID-19). One of the effects of COVID-19 infection is anosmia (i.e., a loss of smell). Unfortunately, this olfactory dysfunction is persistent in around 5% of the world's population, and there is not an effective treatment for it yet. The aim of this paper is to describe a potential non-invasive neurostimulation strategy for treating persistent anosmia in post-COVID-19 patients. In order to design the neurostimulation strategy, 25 subjects who experienced anosmia due to COVID-19 infection underwent an olfactory assessment while their electroencephalographic (EEG) signals were recorded. These signals were used to investigate the activation of brain regions during the olfactory process and identify which regions would be suitable for neurostimulation. Afterwards, 15 subjects participated in the evaluation of the neurostimulation strategy, which was based on applying transcranial direct current stimulation (tDCS) in selected brain regions related to olfactory function. The results showed that subjects with lower scores in the olfactory assessment obtained greater improvement than the other subjects. Thus, tDCS could be a promising option for people who have not fully regained their sense of smell following COVID-19 infection.

An EEG database for the cognitive assessment of motor imagery during walking with a lower-limb exoskeleton.

One important point in the development of a brain-machine Interface (BMI) commanding an exoskeleton is the assessment of the cognitive engagement of the subject during the motor imagery tasks conducted. However, there are not many databases that provide electroencephalography (EEG) data during the use of a lower-limb exoskeleton. The current paper presents a database designed with an experimental protocol aiming to assess not only motor imagery during the control of the device, but also the attention to gait on flat and inclined surfaces. The research was conducted as an EUROBENCH subproject in the facilities sited in Hospital Los Madroños, Brunete (Madrid). The data validation reaches accuracies over 70% in the assessment of motor imagery and attention to gait, which marks the present database as a valuable resource for researches interested on developing and testing new EEG-based BMIs.

Brain-computer interface enhanced by virtual reality training for controlling a lower limb exoskeleton.

This study explores the use of a brain-computer interface (BCI) based on motor imagery (MI) for the control of a lower limb exoskeleton to aid in motor recovery after a neural injury. The BCI was evaluated in ten able-bodied subjects and two patients with spinal cord injuries. Five able-bodied subjects underwent a virtual reality (VR) training session to accelerate training with the BCI. Results from this group were compared with a control group of five able-bodied subjects, and it was found that the employment of shorter training by VR did not reduce the effectiveness of the BCI and even improved it in some cases. Patients gave positive feedback about the system and were able to handle experimental sessions without reaching high levels of physical and mental exertion. These results are promising for the inclusion of BCI in rehabilitation programs, and future research should investigate the potential of the MI-based BCI system.

Brain-machine interface based on transfer-learning for detecting the appearance of obstacles during exoskeleton-assisted walking.

Introduction: Brain-machine interfaces (BMIs) attempt to establish communication between the user and the device to be controlled. BMIs have great challenges to face in order to design a robust control in the real field of application. The artifacts, high volume of training data, and non-stationarity of the signal of EEG-based interfaces are challenges that classical processing techniques do not solve, showing certain shortcomings in the real-time domain. Recent advances in deep-learning techniques open a window of opportunity to solve some of these problems. In this work, an interface able to detect the evoked potential that occurs when a person intends to stop due to the appearance of an unexpected obstacle has been developed. Material and methods: First, the interface was tested on a treadmill with five subjects, in which the user stopped when an obstacle appeared (simulated by a laser). The analysis is based on two consecutive convolutional networks: the first one to discern the intention to stop against normal walking and the second one to correct false detections of the previous one. Results and discussion: The results were superior when using the methodology of the two consecutive networks vs. only the first one in a cross-validation pseudo-online analysis. The false positives per min (FP/min) decreased from 31.8 to 3.9 FP/min and the number of repetitions in which there were no false positives and true positives (TP) improved from 34.9% to 60.3% NOFP/TP. This methodology was tested in a closed-loop experiment with an exoskeleton, in which the brain-machine interface (BMI) detected an obstacle and sent the command to the exoskeleton to stop. This methodology was tested with three healthy subjects, and the online results were 3.8 FP/min and 49.3% NOFP/TP. To make this model feasible for non-able bodied patients with a reduced and manageable time frame, transfer-learning techniques were applied and validated in the previous tests, and were then applied to patients. The results for two incomplete Spinal Cord Injury (iSCI) patients were 37.9% NOFP/TP and 7.7 FP/min.

Riemannian classification analysis for model EEG intention speed patterns.

In this paper, the paradigm of the intention of speed changes from EEG signals with Riemannian classifiers methods is studied in 10 subjects. In addition, the best frequency band and how different electrode configurations affect the accuracy of the model are analyzed. In the prediction of the intention to change speed, results of 68.6% were obtained, in the one of only Increase, results of 64.41 % were obtained, and in the one of only Decrease, results of 71.5% were obtained.

Assessing user experience with BMI-assisted exoskeleton in patients with spinal cord injury.

Spinal Cord Injury (SCI) refers to damage to the spinal cord that can affect different body functionalities. Recovery after SCI depends on multiple factors, being the rehabilitation therapy one of them. New approaches based on robot-assisted training offer the possibility to make training sessions longer and with a reproducible pattern of movements. The control of these robotic devices by means of Brain-Machine Interfaces (BMIs) based on Motor Imagery (MI) favors the patient cognitive engagement during the rehabilitation, promoting mechanisms of neuroplasticity. This research evaluates the acceptance and feedback received from patients with incomplete SCI about the usage of a MI-based BMI with a lower-limb exoskeleton. Clinical Relevance- Patients experienced satisfaction when using the exoskeleton and levels of mental and physical workload were withing reasonable limits. In addition results from the BMI were promising for the inclusion of this type of systems in rehabilitation programs.

Decoding of Turning Intention during Walking Based on EEG Biomarkers.

In the EEG literature, there is a lack of asynchronous intention models that realistically propose interfaces for applications that must operate in real time. In this work, a novel BMI approach to detect in real time the intention to turn is proposed. For this purpose, an offline, pseudo-online and online analysis is presented to validate the EEG as a biomarker for the intention to turn. This article presents a methodology for the creation of a BMI that could differentiate two classes: monotonous walk and intention to turn. A comparison of some of the most popular algorithms in the literature is conducted. To filter the signal, two relevant algorithms are used: H∞ filter and ASR. For processing and classification, the mean of the covariance matrices in the Riemannian space was calculated and then, with various classifiers of different types, the distance of the test samples to each class in the Riemannian space was estimated. This dispenses with power-based models and the necessary baseline correction, which is a problem in realistic scenarios. In the cross-validation for a generic selection (valid for any subject) and a personalized one, the results were, on average, 66.2% and 69.6% with the best filter H∞. For the pseudo-online, the custom configuration for each subject was an average of 40.2% TP and 9.3 FP/min; the best subject obtained 43.9% TP and 2.9 FP/min. In the final validation test, this subject obtained 2.5 FP/min and an accuracy rate of 71.43%, and the turn anticipation was 0.21 s on average.

Quantification of high and low sEMG spectral components during sustained isometric contraction.

Superficial Electromyography (sEMG) spectrum contains aggregated information from several underlying physiological processes. Due to technological limitations, the isolation of these processes is challenging, and therefore, the interpretation of changes in muscle activity frequency is still controversial. Recent studies showed that the spectrum of sEMG signals recorded from isotonic and short-term isometric contractions can be decomposed into independent components whose spectral features recall those of motor unit action potentials. In this paper sEMG spectral decomposition is tested during muscle fatigue induced by long-term isometric contraction where sEMG spectral changes have been widely studied. The main goals of this work are to validate spectral component extraction during long-term isometric muscle activation and the quantification of energy exchange between the low- and high-frequency bands of sEMG signals during muscle fatigue.

Review of tDCS Configurations for Stimulation of the Lower-Limb Area of Motor Cortex and Cerebellum.

This article presents an exhaustive analysis of the works present in the literature pertaining to transcranial direct current stimulation(tDCS) applications. The aim of this work is to analyze the specific characteristics of lower-limb stimulation, identifying the strengths and weaknesses of these works and framing them with the current knowledge of tDCS. The ultimate goal of this work is to propose areas of improvement to create more effective stimulation therapies with less variability.

Detection of the Intention of Direction Changes During Gait Through EEG Signals.

Brain-Computer Interfaces (BCIs) are becoming an important technological tool for the rehabilitation process of patients with locomotor problems, due to their ability to recover the connection between brain and limbs by promoting neural plasticity. They can be used as assistive devices to improve the mobility of handicapped people. For this reason, current BCIs have to be improved to allow an accurate and natural use of external devices. This work proposes a novel methodology for the detection of the intention to change the direction during gait based on event-related desynchronization (ERD). Frequency and temporal features of the electroencephalographic (EEG) signals are characterized. Then, a selection of the most influential features and electrodes to differentiate the direction change intention from the walking is carried out. Best results are obtained when combining frequency and temporal features with an average accuracy of [Formula: see text]%, which are promising to be applied for future BCIs.

Sensory Integration in Human Movement: A New Brain-Machine Interface Based on Gamma Band and Attention Level for Controlling a Lower-Limb Exoskeleton.

Brain-machine interfaces (BMIs) can improve the control of assistance mobility devices making its use more intuitive and natural. In the case of an exoskeleton, they can also help rehabilitation therapies due to the reinforcement of neuro-plasticity through repetitive motor actions and cognitive engagement of the subject. Therefore, the cognitive implication of the user is a key aspect in BMI applications, and it is important to assure that the mental task correlates with the actual motor action. However, the process of walking is usually an autonomous mental task that requires a minimal conscious effort. Consequently, a brain-machine interface focused on the attention to gait could facilitate sensory integration in individuals with neurological impairment through the analysis of voluntary gait will and its repetitive use. This way the combined use of BMI+exoskeleton turns from assistance to restoration. This paper presents a new brain-machine interface based on the decoding of gamma band activity and attention level during motor imagery mental tasks. This work also shows a case study tested in able-bodied subjects prior to a future clinical study, demonstrating that a BMI based on gamma band and attention-level paradigm allows real-time closed-loop control of a Rex exoskeleton.

Selection of Spatial, Temporal and Frequency Features to Detect Direction Changes During Gait.

This paper studies the direction changes during the gait by means of two different distributions of electrodes located in the motor, premotor and occipital areas. The objective is analyzing which areas are involved in the detection of the intention of turning while the person is walking. The signals in both options are characterized with frequency and temporal features and classified following a cross-validation process. A 95% of success rate is achieved when the electrodes are disposed along the motor, premotor and occipital areas.Clinical Relevance- The objective of this study is applying the acknowledgements obtained in the designing of a brain-machine interface (BMI) based in the detection of the intention of the direction change during the gait. This BMI has clinical relevance in the rehabilitation of the gait in patients with motor injuries, assisting the patient to perform the movements as realistic as it is possible.

Usability and acceptance of using a lower-limb exoskeleton controlled by a BMI in incomplete spinal cord injury patients: a case study.

Spinal cord injury (SCI) limits life expectancy and causes a restriction of patient's daily activities. In the last years, robotics exoskeletons have appeared as a promising rehabilitation and assistance tool for patients with motor limitations, as people that have suffered a SCI. The usability and clinical relevance of these robotics systems could be further enhanced by brain-machine interfaces (BMIs), as they can be used to foster patients' neuroplasticity. However, there are not many studies showing the use of BMIs to control exoskeletons with patients. In this work we show a case study where one SCI patient has used a BMI based on motor imagery (MI) in order to control a lower limb exoskeleton that assists their gait.

Analysis of the EEG Rhythms Based on the Empirical Mode Decomposition During Motor Imagery When Using a Lower-Limb Exoskeleton. A Case Study.

The use of brain-machine interfaces in combination with robotic exoskeletons is usually based on the analysis of the changes in power that some brain rhythms experience during a motion event. However, this variation in power is frequently obtained through frequency filtering and power estimation using the Fourier analysis. This paper explores the decomposition of the brain rhythms based on the Empirical Mode Decomposition, as an alternative for the analysis of electroencephalographic (EEG) signals, due to its adaptive capability to the local oscillations of the data, showcasing it as a viable tool for future BMI algorithms based on motor related events.

Study of the Functional Brain Connectivity and Lower-Limb Motor Imagery Performance After Transcranial Direct Current Stimulation.

The use of transcranial direct current stimulation (tDCS) has been related to the improvement of motor and learning tasks. The current research studies the effects of an asymmetric tDCS setup over brain connectivity, when the subject is performing a motor imagery (MI) task during five consecutive days. A brain-computer interface (BCI) based on electroencephalography is simulated in offline analysis to study the effect that tDCS has over different electrode configurations for the BCI. This way, the BCI performance is used as a validation index of the effect of the tDCS setup by the analysis of the classifier accuracy of the experimental sessions. In addition, the relationship between the brain connectivity and the BCI accuracy performance is analyzed. Results indicate that tDCS group, in comparison to the placebo sham group, shows a higher significant number of connectivity interactions in the motor electrodes during MI tasks and an increasing BCI accuracy over the days. However, the asymmetric tDCS setup does not improve the BCI performance of the electrodes in the intended hemisphere.

Pseudo-Online BMI Based on EEG to Detect the Appearance of Sudden Obstacles during Walking.

The aim of this paper is to describe new methods for detecting the appearance of unexpected obstacles during normal gait from EEG signals, improving the accuracy and reducing the false positive rate obtained in previous studies. This way, an exoskeleton for rehabilitation or assistance of people with motor limitations commanded by a Brain-Machine Interface (BMI) could be stopped in case that an obstacle suddenly appears during walking. The EEG data of nine healthy subjects were collected during their normal gait while an obstacle appearance was simulated by the projection of a laser line in a random pattern. Different approaches were considered for selecting the parameters of the BMI: subsets of electrodes, time windows and classifier probabilities, which were based on a linear discriminant analysis (LDA). The pseudo-online results of the BMI for detecting the appearance of obstacles, with an average percentage of 63.9% of accuracy and 2.6 false positives per minute, showed a significant improvement over previous studies.

Improving Real-Time Lower Limb Motor Imagery Detection Using tDCS and an Exoskeleton.

The aim of this work was to test if a novel transcranial direct current stimulation (tDCS) montage boosts the accuracy of lower limb motor imagery (MI) detection by using a real-time brain-machine interface (BMI) based on electroencephalographic (EEG) signals. The tDCS montage designed was composed of two anodes and one cathode: one anode over the right cerebrocerebellum, the other over the motor cortex in Cz, and the cathode over FC2 (using the International 10-10 system). The BMI was designed to detect two MI states: relax and gait MI; and was based on finding the power at the frequency which attained the maximum power difference between the two mental states at each selected EEG electrode. Two different single-blind experiments were conducted, E1 and a pilot test E2. E1 was based on visual cues and feedback and E2 was based on auditory cues and a lower limb exoskeleton as feedback. Twelve subjects participated in E1, while four did so in E2. For both experiments, subjects were separated into two equally-sized groups: sham and active tDCS. The active tDCS group achieved 12.6 and 8.2% higher detection accuracy than the sham group in E1 and E2, respectively, reaching 65 and 81.6% mean detection accuracy in each experiment. The limited results suggest that the exoskeleton (E2) enhanced the detection of the MI tasks with respect to the visual feedback (E1), increasing the accuracy obtained in 16.7 and 21.2% for the active tDCS and sham groups, respectively. Thus, the small pilot study E2 indicates that using an exoskeleton in real-time has the potential of improving the rehabilitation process of cerebrovascular accident (CVA) patients, but larger studies are needed in order to further confirm this claim.

Effects of tDCS on Real-Time BCI Detection of Pedaling Motor Imagery.

The purpose of this work is to strengthen the cortical excitability over the primary motor cortex (M1) and the cerebro-cerebellar pathway by means of a new transcranial direct current stimulation (tDCS) configuration to detect lower limb motor imagery (MI) in real time using two different cognitive neural states: relax and pedaling MI. The anode is located over the primary motor cortex in Cz, and the cathode over the right cerebro-cerebellum. The real-time brain-computer interface (BCI) designed is based on finding, for each electrode selected, the power at the particular frequency where the most difference between the two mental tasks is observed. Electroencephalographic (EEG) electrodes are placed over the brain's premotor area (PM), M1, supplementary motor area (SMA) and primary somatosensory cortex (S1). A single-blind study is carried out, where fourteen healthy subjects are separated into two groups: sham and active tDCS. Each subject is experimented on for five consecutive days. On all days, the results achieved by the active tDCS group were over 60% in real-time detection accuracy, with a five-day average of 62.6%. The sham group eventually reached those levels of accuracy, but it needed three days of training to do so.