Wyniki wyszukiwania - BazTech

1

Modified Atom Search Optimization-based Deep Recurrent Neural Network for epileptic seizure prediction using electroencephalogram signals

Borhade Ratnaprabha Ravindra, Nagmode Manoj S.

Biocybernetics and Biomedical Engineering

|

2020

|

Vol. 40, no. 4

1638--1653

EN

Context and background: Epilepsy is considered as the common neurological disease in the world. Early prediction of epileptic seizure gained great influence on the epileptic patient's life. Epileptic patients suffer from unpredictable conditions that may occur at any moment. Motivation: Various epileptic seizure prediction methods are introduced for accurately predicting the pre-ictal state of human brain, but to determine the discriminative features poses a major challenge in the medical sector. Hypothesis: Develop a technique for epileptic seizure prediction using electroencephalogram signals that detects the epileptic seizure automatically. Method: In this research, an effective optimization algorithm, named Modified Atom Search Optimization-based Deep Recurrent Neural Network is proposed to perform accurate seizure prediction with less computation time. Here, the Deep Recurrent Neural Network classifier per-forms the seizure prediction using various hidden layers associated in the hierarchy layer based on the optimally selected features. The proposed Modified Atom Search Optimization algorithm is designed using the Squirrel Search Algorithm and Atom Search Optimization. It is worth interesting to note that the proposed Modified Atom Search Optimization-based Deep Recurrent Neural Network performed early and accurate seizure prediction using electroencephalogram signals. Result: The analysis of the proposed SASO-based Deep RNN is carried out using CHB-MIT Scalp EEG dataset using the metrics, namely accuracy, sensitivity, and specificity. The proposed algorithm obtained better performance in terms of specificity, accuracy, and sensitivity with the values of 97.536%, 96.545%, and 96.520% by varying training percentage, and 93.736%, 94.128%, and 96.520% by varying K-fold value. Conclusion: The proposed method has significant benefits like, faster convergence rate, easy to implement, low complexity, high speed, and robustness. The weights of the classifier are optimally trained using the proposed algorithm in order to reveal the effectiveness of prediction performance.

2

Personal identification based on brain networks of EEG signals

Kong W., Jiang B., Fan Q., Zhu L., Wei X.

International Journal of Applied Mathematics and Computer Science

|

2018

|

Vol. 28, no. 4

745--757

EN

Personal identification is particularly important in information security. There are numerous advantages of using electroencephalogram (EEG) signals for personal identification, such as uniqueness and anti-deceptiveness. Currently, many researchers focus on single-dataset personal identification, instead of the cross-dataset. In this paper, we propose a method for cross-dataset personal identification based on a brain network of EEG signals. First, brain functional networks are constructed from the phase synchronization values between EEG channels. Then, some attributes of the brain networks including the degree of a node, the clustering coefficient and global efficiency are computed to form a new feature vector. Lastly, we utilize linear discriminant analysis (LDA) to classify the extracted features for personal identification. The performance of the method is quantitatively evaluated on four datasets involving different cognitive tasks: (i) a four-class motor imagery task dataset in BCI Competition IV (2008), (ii) a two-class motor imagery dataset in the BNCI Horizon 2020 project, (iii) a neuromarketing dataset recorded by our laboratory, (iv) a fatigue driving dataset recorded by our laboratory. Empirical results of this paper show that the average identification accuracy of each data set was higher than 0.95 and the best one achieved was 0.99, indicating a promising application in personal identification.

3

Ictal EEG classification based on amplitude and frequency contours of IMFs

Biju K. S., Hakkim H. A., Jibukumar M. G.

Biocybernetics and Biomedical Engineering

|

2017

|

Vol. 37, no. 1

172--183

EN

Electroencephalogram (EEG) signal serves is a powerful tool in epilepsy detection. This study decomposes intrinsic mode functions (IMFs) into amplitude envelope and frequency functions on a time-scale basis using the analytic function of Hilbert transform. IMFs results from the empirical mode decomposition of EEG signals. Features such as energy and entropy parameters were calculated from the amplitude contour of each IMF. Other features, such as interquartile range, mean absolute deviation and standard deviation are also computed for their instantaneous frequencies. Discriminative features were extracted using a large data-base to classify healthy and ictal EEG signals. Normal EEG segments were differentiated from the seizure attack in individual IMF features, multiple features with individual IMF, and individual features with multiple IMFs. Discriminating capability of three Cases was tested. (i) Artificial neural network and (ii) adaptive neuro-fuzzy inference system classification were used to identify EEG segments with seizure attacks. ANOVA was used to analyze statistical performance. Energy and entropy-based features of instantaneous amplitude and standard deviation of instantaneous frequency of IMF2 and IMF1 have 100% accuracy, sensitivity, and specificity. Good performance with a single feature that represents information of the whole data was obtained. The result involved less complicated computation than other time– frequency analysis techniques.

4

Application of MODWT and log-normal distribution model for automatic epilepsy identification

Li M., Chen W., Zhang T.

Biocybernetics and Biomedical Engineering

|

2017

|

Vol. 37, no. 4

679--689

EN

In this paper, a novel approach based on the maximal overlap discrete wavelet transform (MODWT) and log-normal distribution (LND) model was proposed for identifying epileptic seizures from electroencephalogram (EEG) signals. To carry out this study, we explored the potentials of MODWT to decompose the signals into time-frequency sub-bands till sixth level. And demodulation analysis (DA) was investigated to reveal the subtle envelope information from the sub-bands. All obtained coefficients were then used to calculate LDN features, scale parameter (s) and shape parameter (m), which were fed to a random forest classifier (RFC) for classification. Besides, some experiments have been conducted to evaluate the performance of proposed model in the consideration of various wavelet functions as well as feature extractors. The implementation results demonstrated that our proposed technique has yielded remarkable classification performance for all the concerned problems and outperformed the reported methods in terms of the universality. The major finding of this research is that the proposed technique was capable of classifying EEG segments with satisfied accuracy and clinically acceptable computational time. These advantages have make our technique an attractive diagnostic and monitoring tool, which helps doctors in providing better and timely care for the patients.