The purpose of using brain-computer interface (BCI) is to build a bridge between brain and computer for the disable persons, in order to help them to communicate with the outside world. Electroencephalography (EEG) has low signal to noise ratio (SNR), and there exist some problems in the traditional methods for the feature extraction of EEG, such as low classification accuracy, lack of spatial information and huge amounts of features. To solve these problems, we proposed a new method based on time domain, frequency domain and space domain. In this study, independent component analysis (ICA) and wavelet transform were used to extract the temporal, spectral and spatial features from the original EEG signals, and then the extracted features were classified with the method combined support vector machine (SVM) with genetic algorithm (GA). The proposed method displayed a better classification performance, and made the mean accuracy of the Graz datasets in the BCI Competitions of 2003 reach 96%. The classification results showed that the proposed method with the three domains could effectively overcome the drawbacks of the traditional methods based solely on time-frequency domain when the EEG signals were used to describe the characteristics of the brain electrical signals.
By analyzing the mechanism of pronunciation, traditional acoustic parameters, including fundamental frequency, Mel frequency cepstral coefficients (MFCC), linear prediction cepstrum coefficient (LPCC), frequency perturbation, amplitude perturbation, and nonlinear characteristic parameters, including entropy (sample entropy, fuzzy entropy, multi-scale entropy), box-counting dimension, intercept and Hurst, are extracted as feature vectors for identification of pathological voice. Seventy-eight normal voice samples and 73 pathological voice samples for /a/, and 78 normal samples and 80 pathological samples for /i/ are recognized based on support vector machine (SVM). The results showed that compared with traditional acoustic parameters, nonlinear characteristic parameters could be well used to distinguish between healthy and pathological voices, and the recognition rates for /a/ were all higher than those for /i/ except for multi-scale entropy. That is why the /a/ sound data is used widely in related research at home and abroad for obtaining better identification of pathological voices. Adopting multi-scale entropy for /i/ could obtain higher recognition rate than /a/ between healthy and pathological samples, which may provide some useful inspiration for evaluating vocal compensatory function.
We in the present research proposed a classification method that applied infomax independent component analysis (ICA) to respectively extract single modality features of structural magnetic resonance imaging (sMRI) and positron emission tomography (PET). And then we combined these two features by using a method of weight combination. We found that the present method was able to improve the accurate diagnosis of Alzheimer's disease (AD) and mild cognitive impairment (MCI). Compared AD to healthy controls (HC): the study achieved a classification accuracy of 93.75%, with a sensitivity of 100% and a specificity of 87.64%. Compared MCI to HC: classification accuracy was 89.35%, with a sensitivity of 81.85% and a specificity of 99.36%. The experimental results showed that the bi-modality method performed better than the individual modality in comparison to classification accuracy.
The research of sleep staging is not only the basis of diagnosing sleep related diseases, but also the precondition of evaluating sleep quality, and has important clinical significance. In recent years, the research of automatic sleep staging based on computer has become a hotspot and made some achievements. Feature extraction and feature classification are two key technologies in automatic sleep staging system. In order to achieve effective automatic sleep staging, we proposed a new automatic sleep staging method which combines the energy features and least squares support vector machines (LS-SVM). Firstly, we used FIR band-pass filter to extract the energy features of Pz-Oz channel sleep electroencephalogram (EEG) signals, and compared them with those from wavelet packet transform method. Then we designed an LS-SVM classifier to realize the automatic sleep stage classification. The research showed that FIR band-pass filter (with the Kaiser window) performed better than wavelet packet transform (WPT) for energy feature extraction just in terms of the data from the Sleep-EDF Database and the LS-SVM classifier (with the RBF Kernel function) designed was good, and the automatic sleep staging method proposed in this paper was better than many similar methods from other studies with an average accuracy of 88.89% and had a very prosperous application future.
Sleep electroencephalogram (EEG) is an important index in diagnosing sleep disorders and related diseases. Manual sleep staging is time-consuming and often influenced by subjective factors. Existing automatic sleep staging methods have high complexity and a low accuracy rate. A sleep staging method based on support vector machines (SVM) and feature selection using single channel EEG single is proposed in this paper. Thirty-eight features were extracted from the single channel EEG signal. Then based on the feature selection method F-Score's definition, it was extended to multiclass with an added eliminate factor in order to find proper features, which were used as SVM classifier inputs. The eliminate factor was adopted to reduce the negative interaction of features to the result. Research on the F-Score with an added eliminate factor was further accomplished with the data from a standard open source database and the results were compared with none feature selection and standard F-Score feature selection. The results showed that the present method could effectively improve the sleep staging accuracy and reduce the computation time.
The electroencephalogram (EEG) has proved to be a valuable tool in the study of comprehensive conditions whose effects are manifest in the electrical brain activity, and epilepsy is one of such conditions. In the study, multi-scale permutation entropy (MPE) was proposed to describe dynamical characteristics of EEG recordings from epilepsy and healthy subjects, then all the characteristic parameters were forwarded into a support vector machine (SVM) for classification. The classification accuracies of the MPE with SVM were evaluated by a series of experiments. It is indicated that the dynamical characteristics of EEG data with MPE could identify the differences among healthy, inter-ictal and ictal states, and there was a reduction of MPE of EEG from the healthy and inter-ictal state to the ictal state. Experimental results demonstrated that average classification accuracy was 100% by using the MPE as a feature to characterize the healthy and seizure, while 99.58% accuracy was obtained to distinguish the seizure-free and seizure EEG. In addition, the single-scale permutation entropy (PE) at scales 1-5 was put into the SVM for classification at the same time for comparative analysis. The simulation results demonstrated that the proposed method could be a very powerful algorithm for seizure prediction and could have much better performance than the methods based on single scale PE.
Clinic expert information provides important references for residents in need of hospital care. Usually, such information is hidden in the deep web and cannot be directly indexed by search engines. To extract clinic expert information from the deep web, the first challenge is to make a judgment on forms. This paper proposes a novel method based on a domain model, which is a tree structure constructed by the attributes of search interfaces. With this model, search interfaces can be classified to a domain and filled in with domain keywords. Another challenge is to extract information from the returned web pages indexed by search interfaces. To filter the noise information on a web page, a block importance model is proposed. The experiment results indicated that the domain model yielded a precision 10.83% higher than that of the rule-based method, whereas the block importance model yielded an F1 measure 10.5% higher than that of the XPath method.
Poor and monotonous work could easily lead to a decrease of arousal level of the monitoring work personnel. In order to improve the performance of monitoring work, low arousal level needs to be recognized and awakened. We proposed a recognition method of low arousal by the electroencephalogram (EEG) as the object of study to recognize the low arousal level in the vigilance. We used wavelet packet transform to decompose the EEG signal so the EEG rhythms of each component were obtained, and then we calculated the parameters of relative energy and energy ratio of high-low frequency, and constructed the feature vector to monitor low arousal state in the operation. We finally used support vector machine (SVM) to recognize the low arousal state in the simulate operation. The experimental results showed that the method introduced in this article could well distinguish low arousal level from arousal level in the vigilance and it could also get a high recognition rate. Have been compared with other analysis methods, the present method could more effectively recognize low arousal level and provide better technical support for wake-up mechanism of low arousal state.
Due to the sparsity of brain encoding, the neural ensemble signals recorded by microelectrode arrays contain a lot of noise and redundant information, which could reduce the stability and precision of decoding of motion intent. To solve this problem, we proposed a decoding method based on partial least squares (PLS) feature extraction in our study. Firstly, we extracted the features of spike signals using the PLS, and then classified them with support vector machine (SVM) classifier, and decoded them for motion intent. In this study, we decoded neural ensemble signals based on plus-maze test. The results have shown that the proposed method had a better stability and higher decoding accuracy, due to the PLS combined with classification model which overcame the shortcoming of PLS regression that was easily affected by accumulated effect of noise. Meanwhile, the PLS method extracted fewer features with more useful information in comparison with common feature extraction method. The decoding accuracy of real data sets were 93.59%, 84.00% and 83.59%, respectively.
In order to realize sleep staging automatically and conveniently, we used support vector machine (SVM) to analyze the correlation between heart rate variability and sleep stage experimentally. R-R intervals (RRIs) from 33 cases of sleep clinical data of Tianjin Thoracic Hospital were extracted and analyzed by principal component analysis (PCA). The SVM method was used to establish the model and predict the five sleep stages. The prediction accuracy of three-sleep-stage was higher than 80%, in contrast to sleep scoring annotations marked by physiological experts based on electroencephalogram (EEG) golden standard. The result showed that there was a good correlation between heart rate variability and sleep staging. This method is an important supplement to the traditional sleep staging method and has a great value for clinical application.