Non-invasive Brain Signals Research Articles

Whole-Head Noninvasive Brain Signal Measurement System with High Temporal and Spatial Resolution Using Static Magnetic Field Bias to the Brain.

Noninvasive brain signal measurement techniques are crucial for understanding human brain function and brain-machine interface applications. Conventionally, noninvasive brain signal measurement techniques, such as electroencephalography, magnetoencephalography, functional magnetic resonance imaging, and near-infrared spectroscopy, have been developed. However, currently, there is no practical noninvasive technique to measure brain function with high temporal and spatial resolution using one instrument. We developed a novel noninvasive brain signal measurement technique with high temporal and spatial resolution by biasing a static magnetic field emitted from a coil on the head to the brain. In this study, we applied this technique to develop a groundbreaking system for noninvasive whole-head brain function measurement with high spatiotemporal resolution across the entire head. We validated this system by measuring movement-related brain signals evoked by a right index finger extension movement and demonstrated that the proposed system can measure the dynamic activity of brain regions involved in finger movement with high spatiotemporal accuracy over the whole brain.

In-silico testing of new pharmacology for restoring inhibition and human cortical function in depression

Reduced inhibition by somatostatin-expressing interneurons is associated with depression. Administration of positive allosteric modulators of α5 subunit-containing GABAA receptor (α5-PAM) that selectively target this lost inhibition exhibit antidepressant and pro-cognitive effects in rodent models of chronic stress. However, the functional effects of α5-PAM on the human brain in vivo are unknown, and currently cannot be assessed experimentally. We modeled the effects of α5-PAM on tonic inhibition as measured in human neurons, and tested in silico α5-PAM effects on detailed models of human cortical microcircuits in health and depression. We found that α5-PAM effectively recovered impaired cortical processing as quantified by stimulus detection metrics, and also recovered the power spectral density profile of the microcircuit EEG signals. We performed an α5-PAM dose-response and identified simulated EEG biomarker candidates. Our results serve to de-risk and facilitate α5-PAM translation and provide biomarkers in non-invasive brain signals for monitoring target engagement and drug efficacy.

BELT: Bootstrapped EEG-to-language Training by Natural Language Supervision.

Decoding natural language from noninvasive brain signals has been an exciting topic with the potential to expand the applications of brain-computer interface (BCI) systems. However, current methods face limitations in decoding sentences from electroencephalography (EEG) signals. Improving decoding performance requires the development of a more effective encoder for the EEG modality. Nonetheless, learning generalizable EEG representations remains a challenge due to the relatively small scale of existing EEG datasets. In this paper, we propose enhancing the EEG encoder to improve subsequent decoding performance. Specifically, we introduce the discrete Conformer encoder (D-Conformer) to transform EEG signals into discrete representations and bootstrap the learning process by imposing EEG-language alignment from the early training stage. The D-Conformer captures both local and global patterns from EEG signals and discretizes the EEG representation, making the representation more resilient to variations, while early-stage EEG-language alignment mitigates the limitations of small EEG datasets and facilitates the learning of the semantic representations from EEG signals. These enhancements result in improved EEG representations and decoding performance. We conducted extensive experiments and ablation studies to thoroughly evaluate the proposed method. Utilizing the D-Conformer encoder and bootstrapping training strategy, our approach demonstrates superior decoding performance across various tasks, including word-level, sentence-level, and sentiment-level decoding from EEG signals. Specifically, in word-level classification, we show that our encoding method produces more distinctive representations and higher classification performance compared to the EEG encoders from existing methods. At the sentence level, our model outperformed the baseline by 5.45%, achieving a BLEU-1 score of 42.31%. Furthermore, in sentiment classification, our model exceeded the baseline by 14%, achieving a sentiment classification accuracy of 69.3%.

Towards Voice Reconstruction from EEG during Imagined Speech

Translating imagined speech from human brain activity into voice is a challenging and absorbing research issue that can provide new means of human communication via brain signals. Efforts to reconstruct speech from brain activity have shown their potential using invasive measures of spoken speech data, but have faced challenges in reconstructing imagined speech. In this paper, we propose NeuroTalk, which converts non-invasive brain signals of imagined speech into the user's own voice. Our model was trained with spoken speech EEG which was generalized to adapt to the domain of imagined speech, thus allowing natural correspondence between the imagined speech and the voice as a ground truth. In our framework, an automatic speech recognition decoder contributed to decomposing the phonemes of the generated speech, demonstrating the potential of voice reconstruction from unseen words. Our results imply the potential of speech synthesis from human EEG signals, not only from spoken speech but also from the brain signals of imagined speech.

A comparative study of machine learning methods for classifying ERP scalp distribution

Objective. Machine learning (ML) methods are used in different fields for classification and regression purposes with different applications. These methods are also used with various non-invasive brain signals, including Electroencephalography (EEG) signals to detect some patterns in the brain signals. ML methods are considered critical tools for EEG analysis since could overcome some of the limitations in the traditional methods of EEG analysis such as Event-related potentials (ERPs) analysis. The goal of this paper was to apply ML classification methods on ERP scalp distribution to investigate the performance of these methods in identifying numerical information carried in different finger-numeral configurations (FNCs). FNCs in their three forms of montring, counting, and non-canonical counting are used for communication, counting, and doing arithmetic across the world between children and even adults. Studies have shown the relationship between perceptual and semantic processing of FNCs, and neural differences in visually identifying different types of FNCs. Approach. A publicly available 32-channel EEG dataset recorded for 38 participants while they were shown a picture of an FNC (i.e., three categories and four numbers of 1,2,3, and 4) was used. EEG data were pre-processed and ERP scalp distribution of different FNCs was classified across time by six ML methods, including support vector machine, linear discriminant analysis, naïve Bayes, decision tree, K-nearest neighbor, and neural network. The classification was conducted in two conditions: classifying all FNCs together (i.e., 12 classes) and classifying FNCs of each category separately (i.e., 4 classes). Results. The support vector machine had the highest classification accuracy for both conditions. For classifying all FNCs together, the K-nearest neighbor was the next in line; however, the neural network could retrieve numerical information from the FNCs for category-specific classification. Significance. The significance of this study is in exploring the application of multiple ML methods in recognizing numerical information contained in ERP scalp distribution of different finger-numeral configurations.

An Analysis of Intention Detection Strategies to Control Advanced Assistive Technologies at the CYBATHLON.

With the increasing range of functionalities of advanced assistive technologies (AAT), reliable control and initiation of the desired actions become increasingly challenging for users. In this work, we present an analysis of current practices, user preferences, and usability of AAT intention detection strategies based on a survey among participants with disabilities at the CYBATHLON 2020 Global Edition. We collected data from 35 respondents, using devices in various disciplines and levels of technology maturity. We found that conventional, direct inputs such as buttons and joysticks are used by the majority of AAT (71.4%) due to their simplicity and learnability. However, 22 respondents (62.8%) reported a desire for more natural control using muscle or non-invasive brain signals, and 37.1% even reported an openness to invasive strategies for potentially improved control. The usability of the used strategies in terms of the explored attributes (reliability, mental effort, required learning) was mainly perceived positively, whereas no significant difference was observed across intention detection strategies and device types. It can be assumed that the strategies used during the CYBATHLON realistically represent options to control an AAT in a dynamic, physically and mentally demanding environment. Thus, this work underlines the need for carefully considering user needs and preferences for the selection of intention detection strategies in a context of use outside the laboratory.

Imagined word pairs recognition from non-invasive brain signals using Hilbert transform

Imagined word pairs recognition from non-invasive brain signals using Hilbert transform

Disentangling human grasping type from the object's intrinsic properties using low-frequency EEG signals

Grasping movements are known to activate the fronto-parietal brain networks both in human and non-human primates. However, it is unclear if these activations represent properties of the objects or hand postures or both at different stages of the movement. We manipulated the intrinsic properties of the objects and the grasping types in order to create twelve unique combinations of grasping movements and we investigated, in healthy adult humans, the low-frequency time-domain EEG representation of grasping over different stages of the movement. Next, we implemented two multiclass decoders for the grasp type and objects' properties and evaluated them over time. Furthermore, we investigated the similarity between these grasping EEG representations and categorical models that encode properties of the movement and intrinsic properties of the objects. We found that properties of the grasping movement (grasp types, number of fingers) and intrinsic object properties (shape and size) as represented in EEG are encoded in different brain areas throughout the movement stages. Both object properties and grasp types can be decoded significantly above chance level using low-frequency EEG activity during the planning and execution of the movement. Moreover, we found that this preferential time-wise encoding allows the decoding of object properties already from the observation stage, while the grasp type can also be accurately decoded at the object release stage. These findings contribute to the understanding of the grasping representation based on noninvasive EEG brain signals, and its evolution over the course of movement in relation to categorical models that describe the grasped object's properties or that encode properties of the grasping movement. Moreover, our multiclass grasping decoders are informative for the design and implementation of noninvasive motor control strategies.

Movement Related Cortical Potentials in Parkinson's Disease Patients with Freezing of Gait.

Parkinson's disease (PD) is a complex neurode-generative disorder that results into poverty of movements. Freezing of gait (FOG) is a common and severely incapacitating symptom of PD. However, the underlying neurophysiological mechanism of FOG is still unclear. Electroencephalogram (EEG) as a non-invasive brain signal measurement method has been repeatedly used to reveal the pathological reasons behind PD. Since PD affects movement abilities, one particular type of EEG pattern called Movement-related cortical potential (MRCP), which was shown to be related to the generation of movement intention, can be a useful tool in PD research. However, most MRCP studies in PD area focus on the upper-limb movement like finger movement, to the best knowledge of the authors, no study explored MRCP characteristics during lower extremity movement in PD patients, particularly PD patients with FOG. This paper intends to investigate the relationship between the characteristics of MRCP and FOG by comparing MRCP signals among three groups: healthy controls group, PD patients without FOG group, and PD patients with FOG group. These three groups of participants were recruited based on age-matching and gender proportion matching criteria. During the experiment, the Electromyography (EMG) from the tibialis anterior (TA) muscle and EEG were recorded when performing ankle dorsiflexion (AD) by the dominant foot. MRCP signals were extracted from EEG data based on the muscle activities in the TA muscle of the dominant foot. The results of this study show that, during lower limb activities, MRCP shares considerable similarities between PD patients and healthy participants. However, the amplitude of MRCP in PD patients (with and without FOG) is lower compared to the healthy participants. In addition, direct observations from the results show that MRCP from PD patients with FOG has higher amplitude but more variability in terms of amplitude and latency than those from PD patients without FOG.

Characterizing peaks in the EEG power spectrum

The EEG is an important noninvasive brain signal measured from scalp electrodes with the potential for application in the diagnosis of neurological and mental health disorders. In this context analysis of the spectral properties of the EEG are an important aspect of understanding the signal. One characteristic of the signal that has been observed is the presence of intermittent periodicity or regular oscillatory activity under certain conditions. Typically, though not always, this periodicity occurs at ∼10 Hz (called alpha oscillations) and is observed most reliably under eyes closed conditions as a peak in the power spectrum. In the literature the alpha oscillation is generally characterized by the power in the relevant band of the power spectrum. However, since it occurs embedded within a background of nonperiodic activity with a 1/fγ spectral envelope, this approach does not separate the oscillation from the background nonperiodic activity and therefore provides an imprecise view that is largely determined by the spectral decay rather than the embedded periodicity. Here we present an approach to assess periodicity in the EEG signal that is robust to distortions by changes in the 1/f background nonperiodic activity and is particularly suited for detection of small peaks where the band is overwhelmed by the background decay. The method utilizes the second order derivative of the FFT of the signal autocorrelation to define the resulting metric ‘Oscillation Energy’ (or Eα for alpha oscillations). We demonstrate its performance compared to several variants of traditional power spectrum measures with simulated signals and show its application to example EEG signals. We believe a more precise metric to characterize peaks arising from periodicity like this will enable researchers in the field to develop a clearer understanding of this feature of the signal and its relation to behaviors.

Toward EEG-Based Biometric Systems: The Great Potential of Brain-Wave-Based Biometrics

Recognition of individuals based on their unique physiological or behavioral characteristics, called biometric identification or authentication, has received much attention by researchers during the last decade. Among the existing biometric traits, brain signaling has emerged as a powerful candidate due to its highly unique nature, which makes it impossible to steal or mimic. Electroencephalography (EEG), a popular, noninvasive method used for recording brain activity, has recently been identified as an efficient technique to develop biometric systems due to its simplicity, portability, and relatively low cost in comparison to other noninvasive brain signal acquisition models. EEG signals are captured during a relaxed rest state from an individual whose eyes can be closed or open and who can also experience active response states. These signals have been reported to be robust carriers of unique personality traits.

On Similarities and Differences of Invasive and Non-Invasive Electrical Brain Signals in Brain-Computer Interfacing

We perceive that some Brain-Computer Interface (BCI) researchers believe in totally different origins of invasive and non-invasive electrical BCI signals. Based on available literature we argue, however, that although invasive and non-invasive BCI signals are different, the underlying origin of electrical BCIs signals is the same.

Multimodal 2D Brain Computer Interface.

In this work we used multimodal, non-invasive brain signal recording systems, namely Near Infrared Spectroscopy (NIRS), disc electrode electroencephalography (EEG) and tripolar concentric ring electrodes (TCRE) electroencephalography (tEEG). 7 healthy subjects participated in our experiments to control a 2-D Brain Computer Interface (BCI). Four motor imagery task were performed, imagery motion of the left hand, the right hand, both hands and both feet. The signal slope (SS) of the change in oxygenated hemoglobin concentration measured by NIRS was used for feature extraction while the power spectrum density (PSD) of both EEG and tEEG in the frequency band 8-30Hz was used for feature extraction. Linear Discriminant Analysis (LDA) was used to classify different combinations of the aforementioned features. The highest classification accuracy (85.2%) was achieved by using features from all the three brain signals recording modules. The improvement in classification accuracy was highly significant (p = 0.0033) when using the multimodal signals features as compared to pure EEG features.

Noninvasive method to assess the electrical brain activity from rats

This research presents a noninvasive method for the acquisition of brain electrical signal in rat. Was used an electroencephalography (EEG) system developed for bovine and adapted to rats. The bipolar electrode system (needle electrodes) was glued on the surface of the head of the animal without surgical procedures and the other electrode was glued to the tail, as ground. The EEG activity was sampled at 120Hz for an hour. The accuracy and precision of the EEG measurement was performed using Fourier analysis and signal energy. For this, the digital signal was divided into sections successive of 3 seconds and was decomposed into four frequency bands: delta (0.3 to 4Hz), theta (4-8Hz), alpha (8-12Hz) and beta (12-30Hz) and energy (µV²) of the series of time filtered were calculated. The method allowed the acquisition of non-invasive electrical brain signals in conscious rats and their frequency patterns were in agreement with previous studies that used surgical procedures to acquire EEG in rats. This system showed accuracy and precision and will allow further studies on behavior and to investigate the action of drugs on the central nervous system in rats without surgical procedures.

Development and Quantitative Performance Evaluation of a Noninvasive EMG Computer Interface

This paper describes a noninvasive electromyography (EMG) signal-based computer interface and a performance evaluation method based on Fitts' law. The EMG signals induced by volitional wrist movements were acquired from four sites in the lower arm to extract users' intentions, and six classes of wrist movements were distinguished using an artificial neural network. Using the developed interface, a user can move the cursor, click buttons, and type text on a computer. The test setup was built to evaluate the developed interface, and the mouse was tested by five volunteers with intact limbs. The performance of the developed computer interface and the mouse was tested at 1.299 and 7.733 b/s, respectively, and these results were compared with the performance of a commercial noninvasive brain signal interface (0.386 b/s). The results show that the developed interface performed better than the commercial interface, but less satisfactorily than a computer mouse. Although some issues remain to be resolved, the developed EMG interface has the potential to help people with motor disabilities to access computers and Internet environments in a natural and intuitive manner.