Search

Abstract Background Bihemispheric transcranial direct current stimulation (tDCS) of the primary motor cortex (M1) can simultaneously modulate bilateral corticospinal excitability and interhemispheric interaction. However, how tDCS affects subacute stroke recovery remains unclear. We investigated the effects of bihemispheric tDCS on motor recovery in subacute stroke patients. Methods We enrolled subacute inpatients who had first-ever ischemic stroke at subcortical regions and moderate-to-severe baseline Fugl-Meyer Assessment of Upper Extremity (FMA-UE) score 2–56. Participants between 14 and 28 days after stroke were double-blind, randomly assigned (1:1) to receive real (n = 13) or sham (n = 14) bihemispheric tDCS (with ipsilesional M1 anode and contralesional M1 cathode, 20 min, 2 mA) during task practice twice daily for 20 sessions in two weeks. Residual integrity of the ipsilesional corticospinal tract was stratified between groups. The primary efficacy outcome was the change in FMA-UE score from baseline (responder as an increase ≥ 10). The secondary measures included changes in the Action Research Arm Test (ARAT), FMA-Lower Extremity (FMA-LE) and explorative resting-state MRI functional connectivity (FC) of target regions after intervention and three months post-stroke. Results Twenty-seven participants completed the study without significant adverse effects. Nineteen patients (70%) had no recordable baseline motor-evoked potentials (MEP-negative) from the paretic forearm. Compared with the sham group, the real tDCS group showed enhanced improvement of FMA-UE after intervention (p

In recent years, many studies have proposed seizure detection algorithms, but most of them require high computing resources and a large amount of memory, which are difficult to implement in wearable devices. This paper proposes a seizure detection algorithm that uses a small number of features to reduce the memory requirements of the algorithm. During feature extraction, this paper proposes an entropy estimation method that uses bitwise operations instead of logarithmic operations to reduce the algorithm’s demand for computing resources. The experimental results show that the computing time can be reduced by about 81.58%. The seizure detection algorithm in this paper is implemented in an ultra-low power embedded system and performs 7 classification tasks in the Bonn data set to verify the performance of the algorithm. The average classification performance is: Accuracy (97.13%), Specificity (97.57%) and Sensitivity (98.42%). Compared with previous studies, the algorithm of this paper has comparable classification performance, but the proposed algorithm only needs 0.23 seconds from feature extraction to classification result, to the best of our knowledge, which is the seizure detection algorithm with the least computing time currently applied to wearable devices.

Many deep-learning-based seizure detection algorithms have achieved good classification, which usually outperformed traditional machine-learning-based algorithms. However, the hand-engineered features increase the computational complexity and potentially have an ineffectiveness problem for the category. Therefore, this paper proposes a novel end-to-end deep-learning model comprising an inception module and a residual module to analyze the multi-scales of original EEG signals and realize seizure detection without feature extraction. Experiments were conducted and evaluated on the Bonn dataset and the CHB-MIT dataset. In the subject-dependent experiments, our model achieved an average F1-score of 69.34% on the CHB-MIT dataset. In subject-independent experiments, our method achieved an average accuracy of 99.04% on the Bonn dataset and an average F1-score of 37.31% on the CHB-MIT dataset. A series of analyses confirmed that our proposed model has better classification performance and lower computational complexity than existing end-to-end seizure detection models.

Several motor imagery classification methods have been developed and achieve higher accuracy. Machine learning (ML) based algorithms utilizing manually designed features often encounter robustness issues, leading to diminished accuracy. While deep learning (DL) based algorithms exhibit promising accuracy, their extensive computational requirements present challenges in implementing them on portable devices, thereby restricting their practical applications. In this paper, we improve the ML-based algorithm’s feature robustness problems by combining common spatial patterns with Riemannian tangent space mapping, enhancing the algorithm’s feature quality. Furthermore, we introduce a method that utilizes the distance between data points and the SVM hyperplane to compute category scores, thereby enhancing classifier performance. Our experiment uses the BCI Competition IV 2A, BCI Competition III 3A, and a self-recorded dataset for subject-specific experiments to validate the algorithm’s classification performance. Experimental results show that the proposed algorithm achieves the best classification performance, with an accuracy of 78.55%, 83.33%, and 57.44% for BCI Competition IV 2A, BCI Competition III 3A, and the self-recorded dataset. Additionally, to assess the practicality of a real-time portable application, we implemented the proposed algorithm on Raspberry Pi and Jetson Nano, measuring their computation time and peak memory usage. The results demonstrate that our algorithm necessitates only 0.08 to 0.3 seconds of computation time and employs a mere 15MB of memory.

Abstract Background Support vector machines (SVMs) based on brain-wise functional connectivity (FC) have been widely adopted for single-subject prediction of patients with schizophrenia, but most of them had small sample size. This study aimed to evaluate the performance of SVMs based on a large single-site dataset and investigate the effects of demographic homogeneity and training sample size on classification accuracy. Methods The resting functional Magnetic Resonance Imaging (fMRI) dataset comprised 220 patients with schizophrenia and 220 healthy controls. Brain-wise FCs was calculated for each participant and linear SVMs were developed for automatic classification of patients and controls. First, we evaluated the SVMs based on all participants and homogeneous subsamples of men, women, younger (18–30 years), and older (31–50 years) participants by 10-fold nested cross-validation. Then, we hold out a fixed test set of 40 participants (20 patients and 20 controls) and evaluated the SVMs based on incremental training sample sizes (N = 40, 80, …, 400). Results We found that the SVMs based on all participants had accuracy of 85.05%. The SVMs based on male, female, young, and older participants yielded accuracy of 84.66, 81.56, 80.50, and 86.13%, respectively. Although the SVMs based on older subsamples had better performance than those based on all participants, they generalized poorly to younger participants (77.24%). For incremental training sizes, the classification accuracy increased stepwise from 72.6 to 83.3%, with >80% accuracy achieved with sample size >240. Conclusions The findings indicate that SVMs based on a large dataset yield high classification accuracy and establish models using a large sample size with heterogeneous properties are recommended for single subject prediction of schizophrenia.

The motor imagery (MI)-based brain computer interface (BCI) is an intuitive interface that enables users to communicate with external environments through their minds. However, current MI-BCI systems ask naïve subjects to perform unfamiliar MI tasks with simple textual instruction or a visual/auditory cue. The unclear instruction for MI execution not only results in large inter-subject variability in the measured EEG patterns but also causes the difficulty of grouping cross-subject data for big-data training. In this study, we designed an BCI training method in a virtual reality (VR) environment. Subjects wore a head-mounted device (HMD) and executed action observation (AO) concurrently with MI (i.e., AO + MI) in VR environments. EEG signals recorded in AO + MI task were used to train an initial model, and the initial model was continually improved by the provision of EEG data in the following BCI training sessions. We recruited five healthy subjects, and each subject was requested to participate in three kinds of tasks, including an AO + MI task, an MI task, and the task of MI with visual feedback (MI-FB) three times. This study adopted a transformer- based spatial-temporal network (TSTN) to decode the user’s MI intentions. In contrast to other convolutional neural network (CNN) or recurrent neural network (RNN) approaches, the TSTN extracts spatial and temporal features, and applies attention mechanisms along spatial and temporal dimensions to perceive the global dependencies. The mean detection accuracies of TSTN were 0.63, 0.68, 0.75, and 0.77 in the MI, first MI-FB, second MI-FB, and third MI-FB sessions, respectively. This study demonstrated the AO + MI gave an easier way for subjects to conform their imagery actions, and the BCI performance was improved with the continual learning of the MI-FB training process.

This study presents an instantaneous spectrum analysis for electroencephalograph data processing that would facilitate the practice of learning and instruction through real-time measurements of the learner's cognitive load. The instantaneous spectrum analysis is derived from the ensemble empirical mode decomposition which decomposes signals into a gathering of intrinsic mode functions without mode mixing. The multi-marginal Hilbert-Huang spectrum is introduced to estimate frequency contents. As a result, the amplitude of brain rhythms related to the cognitive load can be determined accurately. A model study was performed at first to test the efficacy of the proposed algorithm by comparing with the Fourier based technique, then a prefrontal experiment was conducted to show the advantages of the proposed method. With the higher resolution and more realistic of the proposed method relative to conventional spectrum analysis, more significant features of the signal can be extracted. We believe that the proposed method has the potential to be a substantial technique in electroencephalograph data analysis.

Olfactory dysfunction is related to several clinical neurodegenerative diseases, such as Alzheimer's disease, multiple sclerosis, degenerative ataxias, Parkinson's disease, and so on. Owing to the individual difference in the sensory adaption of human smell function, the olfactory responses usually exhibit large inter-individual difference and change over time after repeated stimulations. The traditional analysis tools, such as statistical parametric mapping (SPM) in functional magnetic resonance imaging technique (fMRI) analysis, utilize the paradigm-based linear correlation and statistical techniques to discriminate the activation areas from background activities. However, these traditional approaches extract olfactory-induced responses using the stereotypic template or paradigm generated model. The olfactory-induced hemodynamic responses affected by internal/external events, such as changes of smell fatigue and attention, are not considered and therefore could result in misleading results. In this paper, owing to the stochastic characteristic of olfactory-induced responses, we adopted multivariate empirical mode decomposition (MEMD) to extract olfactory-induced hemodynamic responses in MRI blood-oxygen-level dependent (BOLD) signals. The MEMD is an efficient data-driven approach to extract nonlinear and non-stationary signals, which is an improved method to expand traditional empirical mode decomposition (EMD) from one channel to multi-channel processing. We applied MEMD to decompose time series of BOLD signals from each slice into multivariate intrinsic mode functions (IMF). The MEMD enables common features of different scales in an image slice to be arranged in distinct IMFs. Each IMF is an analytic, self-constructed, well-defined, and data-driven function with time-varying frequencies. Each IMF was examined by checking its correlation with the paradigm-generated template. The task-related IMFs were chosen to reconstruct olfactory-induced hemodynamic responses. The group analysis of MEMD-processed data showed olfactory-induced activations in the anterior cingulate cortex and middle frontal gyrus.

The sense of smell is one of the most important organs in humans, and olfactory imaging can detect signals in the anterior orbital frontal lobe. This study assessed olfactory stimuli using support vector machines (SVMs) with signals from functional near-infrared spectroscopy (fNIRS) data obtained from the prefrontal cortex. These data included odor stimuli and air state, which triggered the hemodynamic response function (HRF), determined from variations in oxyhemoglobin (oxyHb) and deoxyhemoglobin (deoxyHb) levels; photoplethysmography (PPG) of two wavelengths (raw optical red and near-infrared data); and the ratios of data from two optical datasets. We adopted three SVM kernel functions (i.e., linear, quadratic, and cubic) to analyze signals and compare their performance with the HRF and PPG signals. The results revealed that oxyHb yielded the most efficient single-signal data with a quadratic kernel function, and a combination of HRF and PPG signals yielded the most efficient multi-signal data with the cubic function. Our results revealed superior SVM analysis of HRFs for classifying odor and air status using fNIRS data during olfaction in humans. Furthermore, the olfactory stimulation can be accurately classified by using quadratic and cubic kernel functions in SVM, even for an individual participant data set.

Visual optics for brain-computer interfaces (BCIs) has undergone several developments. Currently, visual optics only provides multiple single-light sources, which limit its applications. This study uses several single-light sources to achieve continuous-phase visual optics. This is combined with a liquid-crystal display (LCD) panel to develop an interactive instrument. Although light synthesis is a nonlinear process, the nonlinear problem is solved by increasing the number of light sources. The measurement data confirm that the phase is approximately linear. This interactive instrument is useful in the biomedical engineering field because it predicts user gaze location.

The longitudinal relationship between central plastic changes and clinical presentations of peripheral hearing impairment remains unknown. Previously, we reported a unique plastic pattern of "healthy-side dominance" in acute unilateral idiopathic sudden sensorineural hearing loss (ISSNHL). This study aimed to explore whether such hemispheric asymmetry bears any prognostic relevance to ISSNHL along the disease course. Using magnetoencephalography (MEG), inter-hemispheric differences in peak dipole amplitude and latency of N100m to monaural tones were evaluated in 21 controls and 21 ISSNHL patients at two stages: initial and fixed stage (1 month later). Dynamics/Prognostication of hemispheric asymmetry were assessed by the interplay between hearing level/hearing gain and ipsilateral/contralateral ratio (I/C) of N100m latency and amplitude. Healthy-side dominance of N100m amplitude was observed in ISSNHL initially. The pattern changed with disease process. There is a strong correlation between the hearing level at the fixed stage and initial I/C(amplitude) on affected-ear stimulation in ISSNHL. The optimal cut-off value with the best prognostication effect for the hearing improvement at the fixed stage was an initial I/C(latency) on affected-ear stimulation of 1.34 (between subgroups of complete and partial recovery) and an initial I/C(latency) on healthy-ear stimulation of 0.76 (between subgroups of partial and no recovery), respectively. This study suggested that a dynamic process of central auditory plasticity can be induced by peripheral lesions. The hemispheric asymmetry at the initial stage bears an excellent prognostic potential for the treatment outcomes and hearing level at the fixed stage in ISSNHL. Our study demonstrated that such brain signature of central auditory plasticity in terms of both N100m latency and amplitude at defined time can serve as a prognostication predictor for ISSNHL. Further studies are needed to explore the long-term temporal scenario of auditory hemispheric asymmetry and to get better psychoacoustic correlates of pathological hemispheric asymmetry in ISSNHL.

BackgroundPsychologically, females are usually thought to be superior in interpersonal sensitivity than males. The human mirror-neuron system is considered to provide the basic mechanism for social cognition. However, whether the human mirror-neuron system exhibits gender differences is not yet clear.Methodology/principal findingsWe measured the electroencephalographic mu rhythm, as a reliable indicator of the human mirror-neuron system activity, when female (N = 20) and male (N = 20) participants watched either hand actions or a moving dot. The display of the hand actions included androgynous, male, and female characteristics. The results demonstrate that females displayed significantly stronger mu suppression than males when watching hand actions. Instead, mu suppression was similar across genders when participants observed the moving dot and between the perceived sex differences (same-sex vs. opposite-sex). In addition, the mu suppressions during the observation of hand actions positively correlated with the personal distress subscale of the interpersonal reactivity index and negatively correlated with the systemizing quotient.Conclusions/significanceThe present findings indirectly lend support to the extreme male brain theory put forward by Baron-Cohen (2005), and may cast some light on the mirror-neuron dysfunction in autism spectrum disorders. The mu rhythm in the human mirror-neuron system can be a potential biomarker of empathic mimicry.