This study presents a non-invasive methodology for recording electroencephalography (EEG) in awake, unrestrained dogs using a compact, wireless 8-channel system. EEG data were collected from two dogs of different breeds during an auditory oddball paradigm without sedation. eight gold cup electrodes were symmetrically positioned on the scalp, and recordings were sampled at 250 Hz. Signal quality was validated using power spectral density analysis. Auditory evoked potentials (AEPs) reliably differentiated target from non-target tones, with a clear N100 response observed. In both dogs, the largest amplitudes were recorded over the right lateral temporal region. These findings demonstrate the feasibility of rapid, stress-free EEG acquisition in dogs and other non-human animals and support the use of mobile neurotechnology in naturalistic settings. This approach advances canine cognitive neuroscience and contributes more broadly to the study of brain activity in ecologically valid conditions across species.

In this study, a few-shot transfer learning approach was introduced to decode movement intention from electroencephalographic (EEG) signals, allowing to recognize new tasks with minimal adaptation. To this end, a dataset of EEG signals recorded during the preparation of complex sub-movements was created from a publicly available data collection. The dataset was divided into two parts: the source domain dataset (including 5 classes) and the support (target domain) dataset, (including 2 classes) with no overlap between the two datasets in terms of classes. The proposed methodology consists in projecting EEG signals into the space-frequency-time domain, in processing such projections (rearranged in channels × frequency frames) by means of a custom EEG-based deep neural network (denoted as EEGframeNET5), and then adapting the system to recognize new tasks through a few-shot transfer learning approach. The proposed method achieved an average accuracy of 72.45 ± 4.19% in the 5-way classification of samples from the source domain dataset, outperforming comparable studies in the literature. In the second phase of the study, a few-shot transfer learning approach was proposed to adapt the neural system and make it able to recognize new tasks in the support dataset. The results demonstrated the system’s ability to adapt and recognize new tasks with an average accuracy of 80 ± 0.12% in discriminating hand opening/closing preparation and outperforming reported results in the literature. This study suggests the effectiveness of EEG in capturing information related to the motor preparation of complex movements, potentially paving the way for BCI systems based on motion planning decoding. The proposed methodology could be straightforwardly extended to advanced EEG signal processing in other scenarios, such as motor imagery or neural disorder classification.

This article investigated the automatic recognition of felt and musically communicated emotions using electroencephalogram (EEG), electrocardiogram (ECG), and acoustic signals, which were recorded from eleven musicians instructed to perform music in order to communicate happiness, sadness, relaxation, and anger. Musicians' self-reports indicated that the emotions they musically expressed were highly consistent with those they actually felt. Results showed that the best classification performances, in a subject-dependent classification using a KNN classifier were achieved by using features derived from both the EEG and ECG (with an accuracy of 98.11%). Which was significantly more accurate than using ECG features alone, but was not significantly more accurate than using EEG features alone. The use of acoustic features alone or in combination with EEG and/or ECG features did not lead to better performances than those achieved with EEG plus ECG or EEG alone. Our results suggest that emotion detection of playing musicians, both felt and musically communicated, when coherent, can be classified in a more reliable way using physiological features than involving acoustic features. The reported machine learning results are a step toward the development of affective brain–computer interfaces capable of automatically inferring the emotions of a playing musician in real-time.

Neurorehabilitation based on Brain-Computer Interfaces (BCIs) show important rehabilitation effects for patients after stroke. Previous studies have shown improvements for patients that are in a chronic stage and/or have severe hemiparesis and are particularly challenging for conventional rehabilitation techniques. Seven stroke patients in chronic phase with lower extremity hemiparesis were recruited. All of them participated in 25 BCI-sessions about 3 times/week. The BCI-system was based on the Motor Imagery (MI) of the paretic-ankle dorsiflexion and healthy-wrist dorsiflexion with Functional Electrical Stimulation (FES) and avatar feedback. Assessments were conducted to assess the changes before and after the therapy. The functional scales used were: 10-meters walking test (10MWT), Range of Motion (ROM) of the ankle dorsiflexion and Timed Up and Go (TUG). Results show a significant increase in the gait speed in the primary measure 10MWT self-velocity of 0.14 m/s (SD = 0.10). This improvement is above of the minimally clinically important difference. The speed in the TUG also significantly increased by 0.08 m/s, P = 0.016. The range of motion also was increased after the therapy, ΔROM passive= 8.83° (SD = 7.22), P=0.016, and ΔROM active = 7.14° (SD = 4.84), P= 0.016. These outcomes show the feasibility of this BCI approach, and further support the growing consensus that these types of tools might develop into a new paradigm for gait rehabilitation tool for stroke patients. However, the results are from seven chronic stroke patients so the authors believe that this approach should be further validated in broader studies involving more patients.

Brain-Computer Interfaces (BCI) show important rehabilitation effects for patients after stroke. Previous studies have also shown improvements for patients that are in a chronic stage and/or have severe hemiparesis, and are particularly challenging for conventional rehabilitation techniques. For this pilot study three stroke patients in chronic phase with hemiparesis in the lower extremity were recruited. BCI system was based on the Motor Imagery (MI) with Functional Electrical Stimulation (FES) and Avatar feedback. The results show improvements in gait and balance measured with 10 Meter Walk Test (10MWT) and Timed Up and Go Test (TUG). Walking speed for 10MWT when walking speed was measured in fast velocity improved in average for 0.16 m/s. Improvements were also measured in ankle dorsiflexion movement ability measured with Range of Motion (ROM). The findings of the current study demonstrate this kind of rehabilitation approach could be effective. However further studies are needed including more patients.

In this publication we present a novel P300 speller that could be used for communication with DOC patients. The stimulation for the P300 is done tactile on seven different body locations. Test on four healthy users have been done and resulted in an average accuracy of 95% when answering simple "yes" versus "no" questions. We further implemented a statistical test, that provides additionally a so called zero-class. If the classification result "yes" or "no" is rather uncertain, and therefore maybe wrong, the speller provides the so called zero-class as output. The answer is therefore considered ambiguous. After adding the zero-class calculations to the analysis of the same data, the percentage of correct answers decreased from 95% to 87.5%, but with the big advantage of no single wrong classification.

Brain-Computer Interfaces (BCI) have proved to be a promising tool for neurorehabilitation. However, BCIs based on conventional methods are not highly accurate and reliable, different brain activity patterns are not optimal for all the users of BCIs and has low information transfer rate. Several studies have shown that the combination of different brain signal acquisition methods can lead to higher performance of BCIs. In this paper, we aim to investigate whether the performance of BCI increases if we combine Electroencephalography (EEG) and functional Near Infrared Spectroscopy (fNIRS) simultaneously for classifying Motor Imagery (MI) tasks of right- versus left-hand grasping movement. The results show enhancement in classification accuracy using a multimodal approach of an EEG + fNIRS BCI with an average increase of approximately 8-10% compared to only EEG-based BCI. This indicates that the hybrid approach in Brain-Computer Interface is capable of enhancing the BCI performance.

Recognition of consciousness using auditory oddball paradigms has become an important research topic in the brain-computer interface (BCI) field. Minimizing the time needed to acquire sufficient data for an assessment could be crucial for patients who have limited concentration. This study aims to reduce the assessment time for auditory oddball paradigms by testing different settings and stimulation approaches. One paradigm uses the subject's own name as deviant sound. The other paradigms use standard sine waves for stimulation. EEG activity was recorded during four different auditory oddball paradigms in a group of nine healthy persons. For comparison, the area under the curve of the P300 of each paradigm was calculated. First, we demonstrate that the name of the subject produced a larger P300 area than the sine tones. More importantly, we found that the name paradigm requires fewer trials to achieve similar results as in a standard auditory paradigm. This means the execution time of the auditory paradigm can be reduced compared to using sine waves.

Brain-computer interfaces (BCIs) have been employed in rehabilitation training for post-stroke patients. In this study, we present the results of the intervention based on BCI triggered functional electrical stimulation (FES) and avatar mirroring. Seven chronic stroke patients participated in 25 sessions of training over 13 weeks. Seven assessments were conducted to observe any behavioral changes before and after the intervention. The primary outcome measure, i.e. the Fugl-Meyer Assessment of the Upper Extremity (FMA-UE), increased significantly by 6.4 points (p=0.048), which is above the minimal clinically important difference (MCID). The Modified Ashworth Scale (MAS), one of the secondary outcome measures, reduced significantly in both the wrist and the finger (p=0.046 and p=0.047 respectively). This study demonstrated motor function improvement and spasticity reduction in chronic stroke patients (n=7) after BCI triggered FES and avatar mirroring. One limitation of this study is that the small sample size may not adequately represent the diverse stroke population. Further work should include a randomized controlled trial to investigate the effectiveness of BCI triggered FES compared to conventional therapies.

Previous neuroimaging studies have reported that the ventral temporal cortex (VTC) processes visual stimuli and thereby establish visual categories, which can be detected in electrophysiological signals such as electrocorticography (ECoG). However, most of the studies are based on visual stimulation through a computer. Thus, the degree to which those categories can be generalized is unclear under real-world conditions. This study extends the findings of a previous experiment, which aimed in real-time detection of visual perception, and investigated whether neural face and kanji categories obtained by computer stimuli can be confirmed in a real-world scenario. The real-time decoder accuracy and latency of two patients with epilepsy revealed that real-world faces and kanji can be detected with 79.9% and 28.4% accuracy, respectively, showing an average online detection latency of 447 ms with respect to presentation time. Hence, the VTC cortex elicits robust and similar responses to computer stimuli and real-world face, leading to a powerful brain-computer interface to track a person's attention in a real-world scenario.

Brain waves contain manifold information about ongoing cognitive processes and related body function. This information, such as power changes in certain frequency bands, can be extracted and interpreted by brain-computer interfaces (BCI). For this publication, we re-evaluated data from two subjects with implanted subdural electrodes who participated in a two-class BCI motor imagery experiment with online feedback. In particular, we compared classification accuracy based on bandpower features extracted from a low-frequency band (LFB, Alpha/Beta, 8-32 Hz) and a high-frequency band (HFB, High-Gamma, 110-140 Hz). The signal processing chain involved bandpower computation, spatial filtering via Common Spatial Patterns (CSP), computing the log-normalized variance, and finally classifying via Linear Discriminant Analysis (LDA). For comparison, we also re-evaluated a comparable motor execution experiment with the same participants. Results of the motor imagery experiments revealed that features derived from the LFB provide consistently higher classification accuracies than features derived from the HFB. In contrast to that, HFB-based features outperform LFB-based features for conventional motor execution experiments.

Patients with disorders of consciousness (DOC) cannot reply to questions or clinical assessments using voluntary motor control, and therefore it is very difficult to assess their cognitive capabilities and conscious awareness. Patients who are locked-in (LIS) are instead fully conscious, and they can communicate with their preserved eye movements. However, when the residual oculomotor activity is also lost (e.g., patients with amyotrophic lateral sclerosis disease of very long duration), the locked-in status becomes complete (CLIS). In CLIS patients, detection of conscious awareness may become very challenging, similarly to the subjects with DOC. mindBEAGLE has a physiological testing battery that uses auditory, vibro-tactile and motor imagery paradigms and brain-computer interface (BCI) technology to assess these patients and even provide communication for some of them. The current study presents results from 5 DOC and 3 LIS patients. The auditory evoked potential (AEP) assessement led to classification accuracies between 0 and 90 %, the vibro-tactile P300 paradigms led to 0 % to 100 % accuracy and the motor imagery paradigms led to accuracies up to 83.3 %. Three of the eight patients could succcessfully establish communication with the mindBEAGLE system. The results show that an assessment battery with auditory, vibro-tactile and motor imagery paradigms is able to identify cognitive functions of DOC and LIS patients. Patients showed substantial fluctuations in EEG measures, assessment results and communication reliability across different days and runs. Therefore, it is important to have a system available that can quickly and easily determine the status of a patient. Successful communication with these patients is also important.

To minimize functional morbidity associated with brain surgery, new preventive approaches (also referred to as "prehabilitation") by using motor-imagery-based computer interfaces (MI-BCIs) can be utilized. To achieve successful MI-BCI performance for prehabilitation purposes, the characteristics of an electrocorticographic (ECoG) signal that is associated with overt motor function ("real movement" - RM) versus covert motor function ("motor imagery" - MI) need to be determined. In our current study, 5 patients with pharmacoresistant epilepsy (2 males, average age 25 years, SD 15), undergoing evaluation for epilepsy surgery participated in both RM and MI tasks. Although the RM- and MI-related ECoG changes had some common features, they also differed in a number of ways, such as location, frequency ranges, signal synchronization and desynchronization. These similarities and differences are discussed in a view of other neuroimaging studies, including magnetoencephalography (MEG) and functional magnetic resonance imaging (fMRI). We emphasize the need for inclusion of a broad spectrum of frequencies in ECoG analysis, when RM- and MI-related activities are concerned.

Brain-Computer Interfaces (BCIs) can provide users with communication, control, and other capabilities based on specific types of brain activity. In the “P300 BCI” approach, a user views a matrix containing letters or other characters and silently counts each time a target item flashes.Classically, characters flashed by briefly reversing color or other basic changes. Recent work has shown that the new "face speller" approach can improve P300 BCI performance. In this approach, each character changes to a human face during each "flash" instead of simply reversing color or other simple changes. The neural activity required to process the attended face, as well as other stimulus changes, may elicit more distinct evoked potentials that can improve classification. While the "face speller" has shown that face presentation may improve BCI performance, it raises the broader possibility that other stimulus changes could provide further performance improvements. The present study explored P300 BCI performance across four conditions. A control condition used "upturned" black and white stimuli that simply reversed color during each flash. Three other conditions explored different face conditions, varying color and the number of different faces presented during each flash. Accuracy was higher in all three face speller conditions than in the conventional "upturned" condition. Colored faces may yield higher accuracy than black and white faces.

Early stages in the development of a Brain-and-Body-Computer Interface controlled robot avatar are presented. The robot is aimed at performing well-defined daily tasks upon the choice and on behalf of a user. We built on recent advances in neuroscience, robotics and machine learning to demonstrate that it is possible to control a robot, accurately and reliably, by decoding scalp-recorded non-invasive Electroencephalographic (EEG) potentials into user intentions.

In this paper a brain-computer interface (BCI) is presented which uses steady-state visual evoked potentials for controlling a robot. EEG is derived from three subjects to test the performance of the system. For feature extraction and classification on one hand the Minimum Energy method, and on the other hand the Fast Fourier Transformation (FFT) with linear discriminant analysis (LDA) is used. As final step a novel method was implemented which analyzes the change rate and the majority weight of redundant classifiers to improve the robustness and to provide a zero classification. The implementation is tested with a robot which is able to move forward, backward, to the left and to the right. High accuracy is achieved for all the commands. Of special interest is, that a zero-class recognition was implemented successfully which causes the robot to stop with high reliability if the subject does not look at one of the stimulation LEDs.