We proposed a new deep learning model by analyzing electroencephalogram signals to reduce the complexity of feature extraction and improve the accuracy of recognition of fatigue status of pilots. For one thing, we applied wavelet packet transform to decompose electroencephalogram signals of pilots to extract the δ wave (0.4–3 Hz), θ wave (4–7 Hz), α wave (8–13 Hz) and β wave (14–30 Hz), and the combination of them was used as de-nosing electroencephalogram signals. For another, we proposed a deep contractive auto-encoding network-Softmax model for identifying pilots' fatigue status. Its recognition results were also compared with other models. The experimental results showed that the proposed deep learning model had a nice recognition, and the accuracy of recognition was up to 91.67%. Therefore, recognition of fatigue status of pilots based on deep contractive auto-encoding network is of great significance.
Seizures during sleep increase the probability of complication and sudden death. Effective prediction of seizures in sleep allows doctors and patients to take timely treatments to reduce the aforementioned probability. Most of the existing methods make use of electroencephalogram (EEG) to predict seizures, which are not specific developed for the sleep. However, EEG during sleep has its characteristics compared with EEG during other states. Therefore, in order to improve the sensitivity and reduce the false alarm rate, this paper utilized the characteristics of EEG to predict seizures during sleep. We firstly constructed the feature vector including the absolute power spectrum, the relative power spectrum and the power spectrum ratio in different frequencies. Secondly, the separation criterion and branch-and-bound method were applied to select features. Finally, support vector machine classifier were trained, which is then employed for online prediction. Compared with the existing method that do not consider the characteristics of sleeping EEG (sensitivity 91.67%, false alarm rate 9.19%), the proposed method was superior in terms of sensitivity (100%) and false alarm rate (2.11%). This method can improve the existing epilepsy prediction methods and has important clinical value.
The brain computer interface (BCI) can be used to control external devices directly through electroencephalogram (EEG) information. A multi-linear principal component analysis (MPCA) framework was used for the limitations of tensor form of multichannel EEG signals processing based on traditional principal component analysis (PCA) and two-dimensional principal component analysis (2DPCA). Based on MPCA, we used the projection of tensor-matrix to achieve the goal of dimensionality reduction and features exaction. Then we used the Fisher linear classifier to classify the features. Furthermore, we used this novel method on the BCI competitionⅡdataset 4 and BCI competitionⅣdataset 3 in the experiment. The second-order tensor representation of time-space EEG data and the third-order tensor representation of time-space-frequency EEG data were used. The best results that were superior to those from other dimensionality reduction methods were obtained by much debugging on parameter P and testQ. For two-order tensor, the highest accuracy rates could be achieved as 81.0% and 40.1%, and for three-order tensor, the highest accuracy rates were 76.0% and 43.5%, respectively.
All the collected original electroencephalograph (EEG) signals were the subjects to low-frequency and spike noise. According to this fact, we in this study performed denoising based on the combination of wavelet transform and independent component analysis (ICA). Then we used three characteristic parameters, complexity, approximate entropy and wavelet entropy values, to calculate the preprocessed EEG data. We then made a distinguishing judge on the EEG state by the state change rate of the characteristic parameters. Through the anesthesia and non-anesthesia EEG data processing results showed that each of the three state change rates could reach about 50.5%, 21.6%, 19.5%, respectively, in which the performance of wavelet entropy was the highest. All of them could be used as a foundation in the quantified research of depth of anesthesia based on EEG analysis.
Sleep staging is the basis for solving sleep problems. There’s an upper limit for the classification accuracy of sleep staging models based on single-channel electroencephalogram (EEG) data and features. To address this problem, this paper proposed an automatic sleep staging model that mixes deep convolutional neural network (DCNN) and bi-directional long short-term memory network (BiLSTM). The model used DCNN to automatically learn the time-frequency domain features of EEG signals, and used BiLSTM to extract the temporal features between the data, fully exploiting the feature information contained in the data to improve the accuracy of automatic sleep staging. At the same time, noise reduction techniques and adaptive synthetic sampling were used to reduce the impact of signal noise and unbalanced data sets on model performance. In this paper, experiments were conducted using the Sleep-European Data Format Database Expanded and the Shanghai Mental Health Center Sleep Database, and achieved an overall accuracy rate of 86.9% and 88.9% respectively. When compared with the basic network model, all the experimental results outperformed the basic network, further demonstrating the validity of this paper's model, which can provide a reference for the construction of a home sleep monitoring system based on single-channel EEG signals.
Rapid and accurate identification and effective non-drug intervention are the worldwide challenges in the field of depression. Electroencephalogram (EEG) signals contain rich quantitative markers of depression, but whole-brain EEG signals acquisition process is too complicated to be applied on a large-scale population. Based on the wearable frontal lobe EEG monitoring device developed by the authors’ laboratory, this study discussed the application of wearable EEG signal in depression recognition and intervention. The technical principle of wearable EEG signals monitoring device and the commonly used wearable EEG devices were introduced. Key technologies for wearable EEG signals-based depression recognition and the existing technical limitations were reviewed and discussed. Finally, a closed-loop brain-computer music interface system for personalized depression intervention was proposed, and the technical challenges were further discussed. This review paper may contribute to the transformation of relevant theories and technologies from basic research to application, and further advance the process of depression screening and personalized intervention.
Aiming at the problem that the unbalanced distribution of data in sleep electroencephalogram(EEG) signals and poor comfort in the process of polysomnography information collection will reduce the model's classification ability, this paper proposed a sleep state recognition method using single-channel EEG signals (WKCNN-LSTM) based on one-dimensional width kernel convolutional neural networks(WKCNN) and long-short-term memory networks (LSTM). Firstly, the wavelet denoising and synthetic minority over-sampling technique-Tomek link (SMOTE-Tomek) algorithm were used to preprocess the original sleep EEG signals. Secondly, one-dimensional sleep EEG signals were used as the input of the model, and WKCNN was used to extract frequency-domain features and suppress high-frequency noise. Then, the LSTM layer was used to learn the time-domain features. Finally, normalized exponential function was used on the full connection layer to realize sleep state. The experimental results showed that the classification accuracy of the one-dimensional WKCNN-LSTM model was 91.80% in this paper, which was better than that of similar studies in recent years, and the model had good generalization ability. This study improved classification accuracy of single-channel sleep EEG signals that can be easily utilized in portable sleep monitoring devices.
Sleep quality is closely related to human health. It is very important to correctly discriminate the sleep stages for evaluating sleep quality, diagnosing and analyzing the sleep-related disorders. Polysomnography (PSG) signals are commonly used to record and analyze sleep stages. Effective feature extraction and representation is one of the most important steps to improve the performance of sleep stage classification. In this work, a collaborative representation (CR) algorithm was adopted to re-represent the original extracted features from electroencephalogram signal, and then the kernel entropy component analysis (KECA) algorithm was further used to reduce the feature dimension of CR-feature. To evaluate the performance of CR-KECA, we compared the original feature, CR feature and readied CR feature (CR-PCA) after principal component analysis (PCA). The experimental results of sleep stage classification indicated that the CR-KECA method achieved the best performance compared with the original feature, CR feature, and CR-PCA feature with the classification accuracy of 68.74±0.46%, sensitivity of 68.76±0.43% and specificity of 92.19±0.11%. Moreover, CR algorithm had low computational complexity, and the feature dimension after KECA was much smaller, which made CR-KECA algorithm suitable for the analysis of large-scale sleep data.
This study uses mind-control game training to intervene in patients with mild cognitive impairment to improve their cognitive function. In this study, electroencephalogram (EEG) data of 40 participants were collected before and after two training sessions. The continuous complexity of EEG signals was analyzed to assess the status of cognitive function and explore the effect of mind-control game training on the improvement of cognitive function. The results showed that after two training sessions, the continuous complexity of EEG signal of the subject increased (0.012 44 ± 0.000 29, P < 0.05) and amplitude of curve fluctuation decreased gradually, indicating that with increase of training times, the continuous complexity increased significantly, the cognitive function of brain improved significantly and state was stable. The results of this paper may show that mind-control game training can improve the status of the brain cognitive function, which may provide support and help for the future intervention of cognitive dysfunction.
As a common disease in nervous system, epilepsy is possessed of characteristics of high incidence, suddenness and recurrent seizures. Timely prediction with corresponding rescues and treatments can be regarded as effective countermeasure to epilepsy emergencies, while most accidental injuries can thus be avoided. Currently, how to use electroencephalogram (EEG) signals to predict seizure is becoming a highlight topic in epilepsy researches. In spite of significant progress that made, more efforts are still to be made before clinical applications. This paper reviews past epilepsy studies, including research records and critical technologies. Contributions of machine learning (ML) and deep learning (DL) on seizure predictions have been emphasized. Since feature selection and model generalization limit prediction ratings of conventional ML measures, DL based seizure predictions predominate future epilepsy studies. Consequently, more exploration may be vitally important for promoting clinical applications of epileptic seizure prediction.