A systematic review on functional electrical stimulation based rehabilitation systems for upper limb post-stroke recovery

Background

Stroke is one of the most common neurological conditions that often leads to upper limb motor impairments, significantly affecting individuals' quality of life. Rehabilitation strategies are crucial in facilitating post-stroke recovery and improving functional independence. Functional Electrical Stimulation (FES) systems have emerged as promising upper limb rehabilitation tools, offering innovative neuromuscular reeducation approaches.

Objective

The main objective of this paper is to provide a comprehensive systematic review of the start-of-the-art functional electrical stimulation (FES) systems for upper limb neurorehabilitation in post-stroke therapy. More specifically, this paper aims to review different types of FES systems, their feasibility testing, or randomized control trials (RCT) studies.

Methods

The FES systems classification is based on the involvement of patient feedback within the FES control, which mainly includes "Open-Loop FES Systems" (manually controlled) and "Closed-Loop FES Systems" (brain-computer interface-BCI and electromyography-EMG controlled). Thus, valuable insights are presented into the technological advantages and effectiveness of Manual FES, EEG-FES, and EMG-FES systems.

Results and discussion

The review analyzed 25 studies and found that the use of FES-based rehabilitation systems resulted in favorable outcomes for the stroke recovery of upper limb functional movements, as measured by the FMA (Fugl-Meyer Assessment) (Manually controlled FES: mean difference = 5.6, 95% CI (3.77, 7.5), P < 0.001; BCI-controlled FES: mean difference = 5.37, 95% CI (4.2, 6.6), P < 0.001; EMG-controlled FES: mean difference = 14.14, 95% CI (11.72, 16.6), P < 0.001) and ARAT (Action Research Arm Test) (EMG-controlled FES: mean difference = 11.9, 95% CI (8.8, 14.9), P < 0.001) scores. Furthermore, the shortcomings, clinical considerations, comparison to non-FES systems, design improvements, and possible future implications are also discussed for improving stroke rehabilitation systems and advancing post-stroke recovery. Thus, summarizing the existing literature, this review paper can help researchers identify areas for further investigation. This can lead to formulating research questions and developing new studies aimed at improving FES systems and their outcomes in upper limb rehabilitation.

Machine Learning for Motor Imagery Wrist Dorsiflexion Prediction in Brain-Computer Interface Assisted Stroke Rehabilitation

Stroke is a life-changing event that can affect the survivors' physical, cognitive and emotional state. Stroke care focuses on helping the survivors to regain their strength; recover as much functionality as possible and return to independent living through rehabilitation therapies. Automated training protocols have been reported to improve the efficiency of the rehabilitation process. These protocols also decrease the dependency of the process on a professional trainer. Brain-Computer Interface (BCI) based systems are examples of such systems where they make use of the motor imagery (MI) based electroencephalogram (EEG) signals to drive the rehabilitation protocols. In this paper, we have proposed the use of well-known machine learning (ML) algorithms, such as, the decision tree (DT), Naive Bayesian (NB), linear discriminant analysis (LDA), support vector machine (SVM), ensemble learning classifier (ELC), and artificial neural network (ANN) for MI wrist dorsiflexion prediction in a BCI assisted stroke rehabilitation study conducted on eleven stroke survivors with either the left or right paresis. The doubling sub-band selection filter bank common spatial pattern (DSBS-FBCSP) has been proposed as feature extractor and it is observed that the ANN based classifier produces the best results.

CACHET-CADB: A Contextualized Ambulatory Electrocardiography Arrhythmia Dataset

ECG is a non-invasive tool for arrhythmia detection. In recent years, wearable ECG-based ambulatory arrhythmia monitoring has gained increasing attention. However, arrhythmia detection algorithms trained on existing public arrhythmia databases show higher FPR when applied to such ambulatory ECG recordings. It is primarily because the existing public databases are relatively clean as they are recorded using clinical-grade ECG devices in controlled clinical environments. They may not represent the signal quality and artifacts present in ambulatory patient-operated ECG. To help build and evaluate arrhythmia detection algorithms that can work on wearable ECG from free-living conditions, we present the design and development of the CACHET-CADB, a multi-site contextualized ECG database from free-living conditions. The CACHET-CADB is subpart of the REAFEL study, which aims at reaching the frail elderly patient to optimize the diagnosis of atrial fibrillation. In contrast to the existing databases, along with the ECG, CACHET-CADB also provides the continuous recording of patients' contextual data such as activities, body positions, movement accelerations, symptoms, stress level, and sleep quality. These contextual data can aid in improving the machine/deep learning-based automated arrhythmia detection algorithms on patient-operated wearable ECG. Currently, CACHET-CADB has 259 days of contextualized ECG recordings from 24 patients and 1,602 manually annotated 10 s heart-rhythm samples. The length of the ECG records in the CACHET-CADB varies from 24 h to 3 weeks. The patient's ambulatory context information (activities, movement acceleration, body position, etc.) is extracted for every 10 s interval cumulatively. From the analysis, nearly 11% of the ECG data in the database is found to be noisy. A software toolkit for the use of the CACHET-CADB is also provided.

Deepaware: A hybrid deep learning and context-aware heuristics-based model for atrial fibrillation detection

BACKGROUND

State-of-the-art automatic atrial fibrillation (AF) detection models trained on RR-interval (RRI) features generally produce high performance on standard benchmark electrocardiogram (ECG) AF datasets. These models, however, result in a significantly high false positive rates (FPRs) when applied on ECG data collected under free-living ambulatory conditions and in the presence of non-AF arrhythmias.

METHOD

This paper proposes DeepAware, a novel hybrid model combining deep learning (DL) and context-aware heuristics (CAH), which reduces the FPR effectively and improves the AF detection performance on participant-operated ambulatory ECG from free-living conditions. It exploits the RRI and P-wave features, as well as the contextual features from the ambulatory ECG.

RESULTS

DeepAware is shown to be very generalizable and superior to the state-of-the-art models when applied on unseen benchmark ECG AF datasets. Most importantly, the model is able to detect AF efficiently when applied on participant-operated ambulatory ECG recordings from free-living conditions and has achieved a sensitivity (Se), specificity (Sp), and accuracy (Acc) of 97.94%, 98.39%, 98.06%, respectively. Results also demonstrate the effect of atrial activity analysis (via P-waves detection) and CAH in reducing the FPR over the RRI features-based AF detection model.

CONCLUSIONS

The proposed DeepAware model can substantially reduce the physician's workload of manually reviewing the false positives (FPs) and facilitate long-term ambulatory monitoring for early detection of AF.

Emergence of flexible technology in developing advanced systems for post-stroke rehabilitation: a comprehensive review

OBJECTIVE

Stroke is one of the most common neural disorders, which causes physical disabilities and motor impairments among its survivors. Several technologies have been developed for providing stroke rehabilitation and to assist the survivors in performing their daily life activities. Currently, the use of flexible technology (FT) for stroke rehabilitation systems is on a rise that allows the development of more compact and lightweight wearable systems, which stroke survivors can easily use for long-term activities.

APPROACH

For stroke applications, FT mainly includes the "flexible/stretchable electronics", "e-textile (electronic textile)" and "soft robotics". Thus, a thorough literature review has been performed to report the practical implementation of FT for post-stroke application.

MAIN RESULTS

In this review, the highlights of the advancement of FT in stroke rehabilitation systems are dealt with. Such systems mainly involve the "biosignal acquisition unit", "rehabilitation devices" and "assistive systems". In terms of biosignals acquisition, electroencephalography (EEG) and electromyography (EMG) are comprehensively described. For rehabilitation/assistive systems, the application of functional electrical stimulation (FES) and robotics units (exoskeleton, orthosis, etc.) have been explained.

SIGNIFICANCE

This is the first review article that compiles the different studies regarding flexible technology based post-stroke systems. Furthermore, the technological advantages, limitations, and possible future implications are also discussed to help improve and advance the flexible systems for the betterment of the stroke community.

Filter bank approach for enhancement of supervised Canonical Correlation Analysis methods for SSVEP-based BCI spellers

Canonical correlation analysis (CCA) is one of the most used algorithms in the steady-state visual evoked potentials (SSVEP)-based brain-computer interface (BCI) systems due to its simplicity, efficiency, and robustness. Researchers have proposed modifications to CCA to improve its speed, allowing high-speed spelling and thus a more natural communication. In this work, we combine two approaches, the filter-bank (FB) approach to extract more information from the harmonics, and a range of different supervised methods which optimize the reference signals to improve the SSVEP detection. The proposed models are tested on the publicly available benchmark dataset for SSVEP-based BCIs and the results show improved performance compared to the state-of-the-art methods and, in particular, the proposed FBMwayCCA approach achieves the best results with an information transfer rate (ITR) of 134.8±8.4 bits/minute. This study indeed suggests the feasibility of combining the fundamental and harmonic SSVEP components with supervised methods in target identification to develop high-speed BCI spellers.

Electromyography and Inertial Motion Sensors Based Wearable Data Acquisition System for Stroke Patients: A Pilot Study

Development of wearable data acquisition systems with applications to human-machine interaction (HMI) is of great interest to assist stroke patients or people with motor disabilities. This paper proposes a hybrid wireless data acquisition system, which combines surface electromyography (sEMG) and inertial measurement unit (IMU) sensors. It is designed to interface wrist extension with external devices, which allows the user to operate devices with hand orientations. A pilot study of the system performed on four healthy subjects has successfully produced two different control signals corresponding to wrist extensions. Preliminary results show a high correlation (0.42-0.75) between sEMG and IMU signals, thus proving the feasibility of such a system. Results also show that the developed system is robust as well as less susceptible to external interferences. The generated control signals can be used to perform real-time control of different devices in daily-life activities, such as turning ON/OFF of lights in a smart home, controlling an electric wheelchair, and other assistive devices. Such a system will help decrease the dependency of disabled people on their caretakers and empower them to perform their daily-life activities independently.

Vocal Imagery vs Intention: Viability of Vocal-Based EEG-BCI Paradigms

The viability of electroencephalogram (EEG) based vocal imagery (VIm) and vocal intention (VInt) Brain-Computer Interface (BCI) systems has been investigated in this study. Four different types of experimental tasks related to humming has been designed and exploited here. They are: (i) non-task specific (NTS), (ii) motor task (MT), (iii) VIm task, and (iv) VInt task. EEG signals from seventeen participants for each of these tasks were recorded from 16 electrode locations on the scalp and its features were extracted and analysed using common spatial pattern (CSP) filter. These features were subsequently fed into a support vector machine (SVM) classifier for classification. This analysis aimed to perform a binary classification, predicting whether the subject was performing one task or the other. Results from an extensive analysis showed a mean classification accuracy of 88.9% for VIm task and 91.1% for VInt task. This study clearly shows that VIm can be classified with ease and is a viable paradigm to integrate in BCIs. Such systems are not only useful for people with speech problems, but in general for people who use BCI systems to help them out in their everyday life, giving them another dimension of system control.

Review on motor imagery based BCI systems for upper limb post-stroke neurorehabilitation: From designing to application

Strokes are a growing cause of mortality and many stroke survivors suffer from motor impairment as well as other types of disabilities in their daily life activities. To treat these sequelae, motor imagery (MI) based brain-computer interface (BCI) systems have shown potential to serve as an effective neurorehabilitation tool for post-stroke rehabilitation therapy. In this review, different MI-BCI based strategies, including "Functional Electric Stimulation, Robotics Assistance and Hybrid Virtual Reality based Models," have been comprehensively reported for upper-limb neurorehabilitation. Each of these approaches have been presented to illustrate the in-depth advantages and challenges of the respective BCI systems. Additionally, the current state-of-the-art and main concerns regarding BCI based post-stroke neurorehabilitation devices have also been discussed. Finally, recommendations for future developments have been proposed while discussing the BCI neurorehabilitation systems.

Motor Imagery based Brain Computer Interface Paradigm for Upper Limb Stroke Rehabilitation

Motor Imagery (MI) based Brain Computer Interface (BCI) systems have shown potential to serve as a tool for neurorehabilitation for post stroke patients to complement the standard therapy. The aim of this study was to develop an MI based BCI system that could potentially be used in neurorehabilitation of hand motor function in stroke patients. Two co-adaptive, three-class MI based BCI systems for realtime processing were developed and compared using the publicly available data from the BCI Competition III Dataset V as well as our own data. The first algorithm utilizes the Filterbank Common Spatial Pattern (FBCSP) for feature extraction, and the other utilizes the Separable Common Spatio-Spectral Pattern (SCSSP) - both combined with a Multi-layer Perceptron (MLP) for classification. The proposed system proved successful when using the competition data showing an average accuracy of 64.71 % for the SCSSP compared to 60.48% for the FBCSP. This proved superior to a related study using the same feature extraction methods, but with other classification methods. The proposed system, however did show results around chance level for the 3-class MI experimental data that we have collected in our laboratory. Further studies needs to be conducted to improve the performance as well as to realize such a system to put in use.

Assessing tDCS Placebo Effects on EEG and Cognitive Tasks

Divergent thinking (DT) using transcranial direct current stimulation (tDCS) has previously been documented with promising results. This paper examines the placebo effect of tDCS. The reaction from a placebo group was tapped using electroencephalogram (EEG). Their performance was measured as a creativity score and compared to a control group. The experiments included multiplication problems and two DT tasks: Alternative Uses Tasks (AUT) and Instances Task (IT). Neither of the groups were sham stimulated during AUT, but during IT the placebo group was sham stimulated. An automatic noise-detection algorithm was developed to remove the speech-induced EEG noise. Features of power, Welchs power spectral density (WPSD) and Welchs cross PSD (WCPSD)/frequency-band/channel were extracted and fed to the Support Vector Machine (SVM) classifiers. The χ²-test showed a significant difference (p<; 0.001) between the no stimulation and sham stimulation conditions when compared to the control group, confirming a placebo effect.

Fast Assessment of Glycemic Control based on Continuous Glucose Monitoring Data

Diabetes has become a major public health problem in the world. In this context, early assessment of glycemic control is essential in order to avoid life-threatening health complications. A panel of diabetes experts have recently proposed a list of recommendations when using Continuous Glucose Monitoring (CGM) for glycemic control assessment including a minimum of two weeks of CGM data. A recent study has further introduced a metric called Glucose Profile Indicator (GPI) for CGM based diabetes management including a subset of the recommended CGM metrics. In this pilot study, it was investigated if less than two weeks of CGM data would impact the performance of GPI compared to the proposed two weeks of CGM data. Furthermore, logistic regression (LR) was used to examine if an improvement could be achieved taking as input the CGM metrics used to quantify GPI. The population mean accuracy for accumulated day 1 to 13 varied between 72.8 ± 2.0% - 98.3 ± 0.4% with no clear sign of improvement using LR. Hence, this indicates a trade-off between the amount of available CGM data and the precision in which the GPI outcome using all 14 days can be achieved when considering features of the GPI alone. Future work is needed to investigate if this trade-off can be improved by the use of additional features of the CGM.

An adaptive CSP filter to investigate user independence in a 3-class MI-BCI paradigm

This paper describes the implementation of a Brain Computer Interface (BCI) scheme using a common spatial patterns (CSP) filter in combination with a Recursive Least Squares (RLS) approach to iteratively update the coefficients of the CSP filter. The proposed adaptive CSP (ACSP) algorithm is made more robust by introducing regularization using Diagonal Loading (DL), and thus will be able to significantly reduce the length of training sessions when introducing new patients to the BCI system. The system is tested on a 4-class multi-limb motor imagery (MI) data set from the BCI competition IV (2a), and a more complex single limb 3-class MI dataset recorded in-house. The latter dataset is produced to mimic an upper limb rehabilitation session, e.g., after stroke. The findings indicate that when extensive calibration data is available, the ACSP performs comparably to the CSP (kappa value of 0.523 and 0.502, respectively, for the 4-class problem); for reduced calibration sessions, the ACSP significantly improved the performance of the system (up to 4-fold). The proposed paradigm proved feasible and the ACSP algorithm seems to enable a user or semi user independent scenario, where the need for long system calibration sessions without feedback is eliminated.

A brain computer interface for robust wheelchair control application based on pseudorandom code modulated Visual Evoked Potential

In this pilot study, a novel and minimalistic Brain Computer Interface (BCI) based wheelchair control application was developed. The system was based on pseudorandom code modulated Visual Evoked Potentials (c-VEPs). The visual stimuli in the scheme were generated based on the Gold code, and the VEPs were recognized and classified using subject-specific algorithms. The system provided the ability of controlling a wheelchair model (LEGO(®) MINDSTORM(®) EV3 robot) in 4 different directions based on the elicited c-VEPs. Ten healthy subjects were evaluated in testing the system where an average accuracy of 97% was achieved. The promising results illustrate the potential of this approach when considering a real wheelchair application.

Automatic Atrial Fibrillation detection: A novel approach using discrete wavelet transform and heart rate variability

Early detection of Atrial Fibrillation (AF) is crucial in order to prevent acute and chronic cardiac rhythm disorders. In this study, a novel method for robust automatic AF detection (AAFD) is proposed by combining atrial activity (AA) and heart rate variability (HRV), which could potentially be used as a screening tool for patients suspected to have AF. The method includes an automatic peak detection prior to the feature extraction, as well as a noise cancellation technique followed by a bagged tree classification. Simulation studies on the MIT-BIH Atrial Fibrillation database was performed to evaluate the performance of the proposed method. Results from these extensive studies showed very promising results, with an average sensitivity of 96.51%, a specificity of 99.19%, and an overall accuracy of 98.22%.

Optimal pseudorandom sequence selection for online c-VEP based BCI control applications

Background

In a c-VEP BCI setting, test subjects can have highly varying performances when different pseudorandom sequences are applied as stimulus, and ideally, multiple codes should be supported. On the other hand, repeating the experiment with many different pseudorandom sequences is a laborious process.

Aims

This study aimed to suggest an efficient method for choosing the optimal stimulus sequence based on a fast test and simple measures to increase the performance and minimize the time consumption for research trials.

Methods

A total of 21 healthy subjects were included in an online wheelchair control task and completed the same task using stimuli based on the m-code, the gold-code, and the Barker-code. Correct/incorrect identification and time consumption were obtained for each identification. Subject-specific templates were characterized and used in a forward-step first-order model to predict the chance of completion and accuracy score.

Results

No specific pseudorandom sequence showed superior accuracy on the group basis. When isolating the individual performances with the highest accuracy, time consumption per identification was not significantly increased. The Accuracy Score aids in predicting what pseudorandom sequence will lead to the best performance using only the templates. The Accuracy Score was higher when the template resembled a delta function the most and when repeated templates were consistent. For completion prediction, only the shape of the template was a significant predictor.

Conclusions

The simple and fast method presented in this study as the Accuracy Score, allows c-VEP based BCI systems to support multiple pseudorandom sequences without increase in trial length. This allows for more personalized BCI systems with better performance to be tested without increased costs.

Steady state visual evoked potential based brain-computer interface for cognitive assessment

Cognitive assessment is of growing importance, with the general population getting older and a rapidly growing incidence of dementia, which is a major public health issue. Treatment of dementia must, to be most effective, start early in the disease process. Thus, early detection of cognitive decline is important. Cognitive decline may be detected using fully-automated computerized assessment. Such systems will provide inexpensive and widely available screenings of cognitive ability. The aim of this pilot study is to develop a real time steady state visual evoked potential (SSVEP) based brain-computer interface (BCI) for neurological cognitive assessment. It is intended for use by patients who suffer from diseases impairing their motor skills, but are still able to control their gaze. Results are based on 11 healthy test subjects. The system performance have an average accuracy of 100% - 0%. The test subjects achieved an information transfer rate (ITR) of 14.64 bits/min - 7.63 bits/min and a subject test performance of 47.22% - 34.10%. This study suggests that BCI may be applicable in practice as a computerized cognitive assessment tool. However, many improvements are required for the system to be fully valid and of clinical use.

A novel approach for automatic detection of Atrial Fibrillation based on Inter Beat Intervals and Support Vector Machine

Atrial fibrillation (AF) is the most common cardiac arrhythmia associated with a major economic burden for the society. Automatic detection of AF in long term recordings can efficiently assist in early diagnosis and management of comorbidities associated with AF. This study presents a novel approach for AF detection based on Inter Beat Intervals (IBI) extracted from long term electrocardiogram (ECG) recordings. Five time-domain features are extracted from the IBIs and a Support Vector Machine (SVM) is used for classification. The results are compared to a state of the art algorithm based on raw ECG. Both algorithms are evaluated on the MIT-BIH Atrial Fibrillation database resulting in equally high classification performance (Sensitivity ≥ 95%). The proposed approach requires detection of R-peaks in the ECG signal but allows for significantly reduced computation time without loss of performance.

Detection of user independent single trial ERPs in Brain Computer Interfaces: An adaptive spatial filtering approach

Brain Computer Interfaces (BCIs) use brain signals to communicate with the external world. The main challenges to address are speed, accuracy and adaptability. Here, a novel algorithm for P300 based BCI spelling system is presented, specifically suited for single-trial detection of Event-Related Potentials (ERPs) by combining spatial filtering and new feature extraction methods. The adaptive spatial filtering technique, axDAWN, removes the need for calibration of the system thereby improving the overall speed of the system. Besides, axDAWN enhances the P300 response to target stimuli. The wavelet decomposition and entropy of the recorded ERPs are shown to be correlated with the presence of the P300 responses. The proposed scheme is validated thoroughly in a P300 speller and provides a solution to achieve high accuracy results for single-trial detection of ERPs, being the system user independent.

A 3D learning playground for potential attention training in ADHD: A brain computer interface approach

This paper presents a novel brain-computer-interface (BCI) system that could potentially be used for enhancing the attention ability of subjects with attention deficit hyperactivity disorder (ADHD). It employs the steady state visual evoked potential (SSVEP) paradigm. The developed system consists of a 3D classroom environment with active 3D distractions and 2D games executed on the blackboard. The system is concealed as a game (with stages of varying difficulty) with an underlying story to motivate the subjects. It was tested on eleven healthy subjects and the results undeniably establish that by moving to a higher stage in the game where the 2D environment is changed to 3D along with the added 3D distractions, the difficulty level in keeping attention on the main task increases for the subjects. Results also show a mean accuracy of 92.26 ± 7.97% and a mean average selection time of 3.07 ± 1.09 seconds.

Neurofeedback Therapy for Enhancing Visual Attention: State-of-the-Art and Challenges

We have witnessed a rapid development of brain-computer interfaces (BCIs) linking the brain to external devices. BCIs can be utilized to treat neurological conditions and even to augment brain functions. BCIs offer a promising treatment for mental disorders, including disorders of attention. Here we review the current state of the art and challenges of attention-based BCIs, with a focus on visual attention. Attention-based BCIs utilize electroencephalograms (EEGs) or other recording techniques to generate neurofeedback, which patients use to improve their attention, a complex cognitive function. Although progress has been made in the studies of neural mechanisms of attention, extraction of attention-related neural signals needed for BCI operations is a difficult problem. To attain good BCI performance, it is important to select the features of neural activity that represent attentional signals. BCI decoding of attention-related activity may be hindered by the presence of different neural signals. Therefore, BCI accuracy can be improved by signal processing algorithms that dissociate signals of interest from irrelevant activities. Notwithstanding recent progress, optimal processing of attentional neural signals remains a fundamental challenge for the development of efficient therapies for disorders of attention.

A comparative study of pseudorandom sequences used in a c-VEP based BCI for online wheelchair control

In this study, a c-VEP based BCI system was developed to run on three distinctive pseudorandom sequences, namely the m-code, the Gold-code, and the Barker-code. The Visual Evoked Potentials (VEPs) were provoked using these codes. In the online session, subjects controlled a LEGO^® Mindstorms^® robot around a fixed track. Choosing the optimal code proved a significant increase in accuracy (p<;0.00001) over the average performance. No single code proved significantly more accurate than the others (p=0.81), suggesting that the term "optimal code" is subject-dependent. However, the Gold-code was significantly faster than both alternatives (p=0.006, p=0.016). When choosing the optimal code for accuracy, no significant decrease in Time Per Identification (TPI) was found (p=0.67). Thus, when creating an online c-VEP based BCI system, it is recommended to use multiple random sequences for increased performance.

Brain-computer interface using P300 and virtual reality: a gaming approach for treating ADHD

This paper presents a novel brain-computer interface (BCI) system aiming at the rehabilitation of attention-deficit/hyperactive disorder in children. It uses the P300 potential in a series of feedback games to improve the subjects' attention. We applied a support vector machine (SVM) using temporal and template-based features to detect these P300 responses. In an experimental setup using five subjects, an average error below 30% was achieved. To make it more challenging the BCI system has been embedded inside an immersive 3D virtual reality (VR) classroom with simulated distractions, which was created by combining a low-cost infrared camera and an "off-axis perspective projection" algorithm. This system is intended for kids by operating with four electrodes, as well as a non-intrusive VR setting. With the promising results, and considering the simplicity of the scheme, we hope to encourage future studies to adapt the techniques presented in this study.

DTU BCI speller: an SSVEP-based spelling system with dictionary support

In this paper, a new brain computer interface (BCI) speller, named DTU BCI speller, is introduced. It is based on the steady-state visual evoked potential (SSVEP) and features dictionary support. The system focuses on simplicity and user friendliness by using a single electrode for the signal acquisition and displays stimuli on a liquid crystal display (LCD). Nine healthy subjects participated in writing full sentences after a five minutes introduction to the system, and obtained an information transfer rate (ITR) of 21.94 ± 15.63 bits/min. The average amount of characters written per minute (CPM) is 4.90 ± 3.84 with a best case of 8.74 CPM. All subjects reported systematically on different user friendliness measures, and the overall results indicated the potentials of the DTU BCI Speller system. For subjects with high classification accuracies, the introduced dictionary approach greatly reduced the time it took to write full sentences.

ECG De-noising: A comparison between EEMD-BLMS and DWT-NN algorithms

Electrocardiogram (ECG) is a widely used non-invasive method to study the rhythmic activity of the heart and thereby to detect the abnormalities. However, these signals are often obscured by artifacts from various sources and minimization of these artifacts are of paramount important. This paper proposes two adaptive techniques, namely the EEMD-BLMS (Ensemble Empirical Mode Decomposition in conjunction with the Block Least Mean Square algorithm) and DWT-NN (Discrete Wavelet Transform followed by Neural Network) methods in minimizing the artifacts from recorded ECG signals, and compares their performance. These methods were first compared on two types of simulated noise corrupted ECG signals: Type-I (desired ECG+noise frequencies outside the ECG frequency band) and Type-II (ECG+noise frequencies both inside and outside the ECG frequency band). Subsequently, they were tested on real ECG recordings. Results clearly show that both the methods works equally well when used on Type-I signals. However, on Type-II signals the DWT-NN performed better. In the case of real ECG data, though both methods performed similar, the DWT-NN method was a slightly better in terms of minimizing the high frequency artifacts.

Semi-supervised adaptation in ssvep-based brain-computer interface using tri-training

This paper presents a novel and computationally simple tri-training based semi-supervised steady-state visual evoked potential (SSVEP)-based brain-computer interface (BCI). It is implemented with autocorrelation-based features and a Naïve-Bayes classifier (NBC). The system uses nine characters presented on a 100 Hz CRT-monitor, three scalp electrodes for signal acquisition, a gUSB-amp for preamplification and two PCs for data-processing and stimulus control respectively. Preliminary test results of the system on nine healthy subjects, with and without tri-training, indicates that the accuracy improves as a result of tri-training.

SSVEP-modulation by covert and overt attention: Novel features for BCI in attention neuro-rehabilitation

In this pilot study the effect of attention (covert and overt) on the signal detection and classification of steady-state visual-evoked potential (SSVEP) were investigated. Using the SSVEP-based paradigm, data were acquired from 4 subjects using 3 scalp electroencephalography (EEG) electrodes located on the visual area. Subjects were instructed to perform the attention task in which they attended covertly or overtly to either of the stimuli flickering with different frequencies (6, 7, 8 and 9Hz). We observed a decrease in signal power in covert compared to the overt attention. However, there was a consistent pattern in covert attention causing an increase in the power of the 2(nd) harmonic of the attended frequency. Encouraging results of this preliminary study indicates that it can be adapted and implemented in the brain-computer interface (BCI) system which could potentially be used as a neuro-rehabilitation tool for individuals with attention deficit.

BCI using imaginary movements: the simulator

Over the past two decades, much progress has been made in the rapidly evolving field of Brain Computer Interface (BCI). This paper presents a novel concept: a BCI-simulator, which has been developed for the Hex-O-Spell interface, using the sensory motor rhythms (SMR) paradigm. With the simulator, it is possible to evaluate how the model parameters such as error classifications, delay between classifications and success rate affect the communication rate. Another advantage of the simulator is that it allows us to study for more classes than most online BCI systems which are limited to only two classes. Results show that the BCI simulator is able to give a deeper understanding of the feedback systems. We also find that a 3-class system is more efficient than a 2-class system if it obtains a success rate of at least 55% of the 2-class system.

An asynchronous P300 BCI with SSVEP-based control state detection

In this paper, an asynchronous brain-computer interface (BCI) system combining the P300 and steady-state visually evoked potentials (SSVEPs) paradigms is proposed. The information transfer is accomplished using P300 event-related potential paradigm and the control state (CS) detection is achieved using SSVEP, overlaid on the P300 base system. Offline and online experiments have been performed with ten subjects to validate the proposed system. It is shown to achieve fast and accurate CS detection without significantly compromising the performance. In online experiments, the system is found to be capable of achieving an average data transfer rate of 19.05 bits/min, with CS detection accuracy of about 88%.

Adaptation in P300 brain-computer interfaces: a two-classifier cotraining approach

A cotraining-based approach is introduced for constructing high-performance classifiers for P300-based brain-computer interfaces (BCIs), which were trained from very little data. It uses two classifiers: Fisher's linear discriminant analysis and Bayesian linear discriminant analysis progressively teaching each other to build a final classifier, which is robust and able to learn effectively from unlabeled data. Detailed analysis of the performance is carried out through extensive cross-validations, and it is shown that the proposed approach is able to build high-performance classifiers from just a few minutes of labeled data and by making efficient use of unlabeled data. An average bit rate of more than 37 bits/min was achieved with just one and a half minutes of training, achieving an increase of about 17 bits/min compared to the fully supervised classification in one of the configurations. This performance improvement is shown to be even more significant in cases where the training data as well as the number of trials that are averaged for detection of a character is low, both of which are desired operational characteristics of a practical BCI system. Moreover, the proposed method outperforms the self-training-based approaches where the confident predictions of a classifier is used to retrain itself.

EOG artifact minimization using oblique projection corrected eigenvector decomposition

In this paper, the authors propose an efficient algorithm to minimize the electrooculogram (EOG) artifacts in electroencephalogram (EEG). The approach uses the eigenvectors obtained from a learning process to initialize an oblique projection based blind source extraction (BSE) algorithm. It is used to extract the point source EOG artifacts. EEG data is subsequently reconstructed by a deflation method. The simulations with synthetic data illustrate that the BSE corrected algorithm is reliable and has better performance than the uncorrected eigenvector decomposition based method. The results of simulations with real EEG data confirms the effectiveness of our algorithm.

fMRI data analysis with nonstationary noise models: a Bayesian approach

The assumption of noise stationarity in the functional magnetic resonance imaging (fMRI) data analysis may lead to the loss of crucial dynamic features of the data and thus result in inaccurate activation detection. In this paper, a Bayesian approach is proposed to analyze the fMRI data with two nonstationary noise models (the time-varying variance noise model and the fractional noise model). The covariance matrices of the time-varying variance noise and the fractional noise after wavelet transform are diagonal matrices. This property is investigated under the Bayesian framework. The Bayesian estimator not only gives an accurate estimate of the weights in general linear model, but also provides posterior probability of activation in a voxel and, hence, avoids the limitations (i.e., using only hypothesis testing) in the classical methods. The performance of the proposed Bayesian methods (under the assumption of different noise models) are compared with the ordinary least squares (OLS) and the weighted least squares (WLS) methods. Results from the simulation studies validate the superiority of the proposed approach to the OLS and WLS methods considering the complex noise structures in the fMRI data.

Estimation of the hemodynamic response of fMRI Data using RBF neural network

Functional magnetic resonance imaging (fMRI) is an important technique for neuroimaging. The conventional system identification methods used in fMRI data analysis assume a linear time-invariant system with the impulse response described by the hemodynamic responses (HDR). However, the measured blood oxygenation level-dependent (BOLD) signals to a particular processing task (for example, rapid event-related fMRI design) show nonlinear properties and vary with different brain regions and subjects. In this paper, radial basis function (RBF) neural network (a powerful technique for modelling nonlinearities) is proposed to model the dynamics underlying the fMRI data. The equivalence of the proposed method to the existing Volterra series method has been demonstrated. It is shown that the first- and second-order Volterra kernels could be deduced from the parameters of the RBF neural network. Studies on both simulated (using Balloon model) as well as real event-related fMRI data show that the proposed method can accurately estimate the HDR of the brain and capture the variations of the HDRs as a function of the brain regions and subjects.

Extraction of fetal electrocardiogram using H ∞ adaptive algorithms

The fetal electrocardiogram (fECG) contains important information regarding the health of the fetus. However, the fECG obtained noninvasively from the abdominal surface electrical recordings of a pregnant woman are dominated by strong interference from the maternal electrocardiogram (mECG). In this paper, based on the H(infinity) principle, two adaptive algorithms are proposed for the extraction of fECG from the trans-abdominal recordings of pregnant women. The motivation behind the application of H(infinity) techniques is the fact that they are robust with respect to model uncertainties and lack of statistical information regarding noise. The proposed algorithms are applied to simulated as well as real multichannel ECG recordings and their performances are compared to that of the well-known least-mean-square (LMS) adaptive algorithm. It is found that the proposed H(infinity) based algorithms perform superior to the LMS algorithm in extracting the fECG signal.

Cardiac State Diagnosis using Adaptive Neuro-Fuzzy Technique

Heart rate signals may either contain indicators of a current disease or even warnings about impending diseases. However, to manually study and pinpoint heart abnormalities in voluminous data is strenuous and time consuming. Here, an adaptive neuro-fuzzy network is used to classify heart abnormalities in ten different cardiac states and shown to be effective.

Bayesian radial basis function network for modeling fMRI data

Noisy and nonlinear nature make fMRI signal processing a challenging problem. We proposed and analyzed the Bayesian trained radial basis function (RBF) neural network in fMRI data processing. The method, which determines the regularization parameter in RBF network automatically by Bayesian learning, is especially suitable for fMRI data processing. Both simulated and real fMRI data were tested. Results show that this approach could model fMRI signals and remove the slowly varying drift in the data sets as well.

Nonlinear analysis of EMG signals - a chaotic approach

This paper aims to present a systematic characterisation of the electromyogram (EMG) signal using a nonlinear chaotic approach. EMG signals from 10 muscles in the leg during walking and maximum voluntary contraction (MVC) were obtained and pre-processed using wavelet based denoising techniques. All signals were tested for non-linearity, stationarity and determinism. Chaotic characterization was done by calculating invariants such as correlation dimension (D2), Lyapunov spectrum (lambda1) and Kaplan-Yorke dimension (D(KY)). The EMG signals were non-linear and short-term stationary. Determinism and structure was found in the phase-space by studying the recurrence plots. Based on the values of the chaotic invariants, EMG signals were found to exhibit signs of chaotic behaviour with a dimension between 2 and 3 for walking and 3 and 4 for MVC data.

Embedding EEG data on 3D head modeled using MRI data

Current 3D head modeling techniques from magnetic resonance imaging (MRI) data fall short of what is needed by clinicians and scientists to study the head anatomy both in terms of accuracy and speed of rendering. This work proposes a novel direct modeling approach (DMA), which successfully and accurately models the 3D head with minimum processing requirements. A method for embedding the electroencephalogram (EEG) dipole source location into this modeled head is also demonstrated. The infomax algorithm is used for performing the independent component analysis (ICA) to obtain temporally independent stationary sources from multichannel EEG data. The dipole source location is then found by using the downhill simplex algorithm on the activation maps of each independent source. Finally a transformation matrix is derived that maps the dipole location onto the 3D head.

Multiuser receiver for DS-CDMA signals in multipath channels: an enhanced multisurface method

This paper deals with the problem of multiuser detection in direct-sequence code-division multiple-access (DS-CDMA) systems in multipath environments. The existing multiuser detectors can be divided into two categories: (1) low-complexity poor-performance linear detectors and (2) high-complexity good-performance nonlinear detectors. In particular, in channels where the orthogonality of the code sequences is destroyed by multipath, detectors with linear complexity perform much worse than the nonlinear detectors. In this paper, we propose an enhanced multisurface method (EMSM) for multiuser detection in multipath channels. EMSM is an intermediate piecewise linear detection scheme with a run-time complexity linear in the number of users. Its bit error rate performance is compared with existing linear detectors, a nonlinear radial basis function detector trained by the new support vector learning algorithm, and Verdu's optimal detector. Simulations in multipath channels, for both synchronous and asynchronous cases, indicate that it always outperforms all other linear detectors, performing nearly as well as nonlinear detectors.

Spatio-temporal modeling and analysis of fMRI data using NARX neural network

This paper presents spatio-temporal modeling and analysis methods to fMRI data. Based on the nonlinear autoregressive with exogenous inputs (NARX) model realized by the Bayesian radial basis function (RBF) neural networks, two methods (NARX-1 and NARX-2) are proposed to capture the unknown complex dynamics of the brain activities. Simulation results on both synthetic and real fMRI data clearly show that the proposed schemes outperform the conventional t-test method in detecting the activated regions of the brain.

Error criteria for cross validation in the context of chaotic time series prediction

The prediction of a chaotic time series over a long horizon is commonly done by iterating one-step-ahead prediction. Prediction can be implemented using machine learning methods, such as radial basis function networks. Typically, cross validation is used to select prediction models based on mean squared error. The bias-variance dilemma dictates that there is an inevitable tradeoff between bias and variance. However, invariants of chaotic systems are unchanged by linear transformations; thus, the bias component may be irrelevant to model selection in the context of chaotic time series prediction. Hence, the use of error variance for model selection, instead of mean squared error, is examined. Clipping is introduced, as a simple way to stabilize iterated predictions. It is shown that using the error variance for model selection, in combination with clipping, may result in better models.

Chaotic synchronization: a nonlinear predictive filtering approach

The synchronization of chaotic systems is a difficult task due to their sensitive dependence on the initial conditions. Perfect synchronization is almost impossible when noise is present in the system. One of the well known stochastic filtering algorithms that is used to synchronize chaotic systems in the presence of noise is the extended Kalman filter (EKF). However, for highly nonlinear systems, the approximation error introduced by the EKF has been shown to be relatively high. In this paper, a nonlinear predictive filter (NPF) is proposed for synchronizing chaotic systems. In this scheme, it is not required to approximate the underlying nonlinearity and hence there is no need to compute the Jacobian of the chaotic system. Numerical simulations are carried out to compare the performances of the NPF and EKF algorithms for synchronizing different sets of chaotic systems and/or maps. The well known Lorenz and Mackey-Glass systems as well as Ikeda map are used for numerical evaluation of the performance. Results clearly show that the NPF based approach is superior to the EKF based approach in terms of the normalized mean square error (NMSE), total NMSE, and the time taken for synchronization (measured in terms of the normalized instantaneous square error) for all the systems and/or maps considered.

Postprocessing methods for finding the embedding dimension of chaotic time series

One problem when using the global false nearest-neighbors (GFNN) method and Cao's method to estimate embedding dimension is that their effectiveness is affected by the ratio of signal power to noise power (SNR). Simple models are proposed to explain the curves commonly obtained when using the GFNN method and Cao's method. Methods are proposed for systematically estimating the embedding dimension. Prior information is incorporated to improve the estimates.

Chaotic dynamics of sea clutter

The notion that a deterministic nonlinear dynamical system (with relatively few degrees of freedom) can display aperiodic behavior has a strong bearing on sea clutter characterization: random-looking sea clutter may be the outcome of a chaotic process. This new approach envisages deterministic rules for the underlying sea clutter dynamics, in contrast to the stochastic approach where sea clutter is viewed as a random process with a large number of degrees of freedom. In this paper, we demonstrate, convincingly for the first time, the chaotic dynamics of sea clutter. We say so on the basis of results obtained using radar data collected from a series of extensive and thorough experiments, which have been carried out with ground-truthed sea clutter data sets at three different sites. The study includes correlation dimension analysis (based on the maximum likelihood principle) and Lyapunov spectrum analysis. The Lyapunov (Kaplan-Yorke) dimension, which is a byproduct of Lyapunov spectrum analysis, shows that it is indeed a good estimator of the correlation dimension. The Lyapunov spectrum also reveals that sea clutter is produced by a coupled system of nonlinear differential equations of order five or six. (c) 1997 American Institute of Physics.