Objective Epilepsy is one of the most common severe brain disorders. Ultrasound deep brain stimulation (UDBS) has shown a potential capability to suppress seizures. However, because seizures occur sporadically, it is necessary to develop a closed-loop system to suppress them. Therefore, we developed a closed-loop wearable UDBS system that delivers ultrasound to the hippocampus to suppress epileptic seizures. Approach Mice were intraperitoneally injected with 10 mg/kg kainic acid (KA) and divided into sham and UDBS groups. Epileptic seizures were detected by applying both long short-term memory (LSTM) and bidirectional LSTM (BILSTM) networks according to EEG signal characteristics. When epileptic seizures were detected, the closed-loop UDBS system automatically activated a trigger switch to stimulate the hippocampus for 10 min and continuously record EEG signals until 20 min after ultrasonic stimulation. EEG signals were analyzed using the MATLAB software. After EEG recording, we observed the survival rate of the experimental mice for 72 hours. Main results The BiLSTM network was found to have preferable classification performance over the LSTM network. The closed-loop UDBS system with BiLSTM could automatically detect epileptic seizures using EEG signals and effectively reduce epileptic EEG power spectral density (PSD) and seizure duration by 10.73%, eventually improving the survival rate of early epileptic mice from 67.57% in the sham group to 88.89% in the UDBS group. Significance The closed-loop UDBS system developed in this study could be an effective clinical tool for the control of epilepsy.

Closed-loop wearable ultrasound deep brain stimulation system based on EEG in mice.

Epilepsy affects about 1 % of the contemporary population and sternly reduces the wellbeing of its patients. It is a neurological disorder of the central nervous system that is usually characterized by sudden seizure. The possibility of detecting and predicting epileptic seizure has engrossed mankind already for over 35 years. One of the main tools in detecting and predicting the Epilepsy seizures are the Electroencephalograms (EEG), which record the brain activity by measuring the extracellular field potentials due to neuronal discharges. This EEG is quite difficult and complex to interpret even by an expert neurologist, even so, it is time-consuming, often challenging, sets in human error as well as delay in treatment. In this research, a hybrid classification model using Grasshopper Optimization Algorithm (GOA) and Artificial Neural Network (ANN) for automatic seizure detection in EEG is proposed called GOA-ANN approach. Nine parameters (mean value, variance value, Standard deviation value, energy value, entropy value and maximum value, RMS value, kurtosis and skewness) were extracted and used as the features to train the ANN classifiers. GOA was used for selecting the best features in order to obtain an effective EEG classification. In comparison with other research, the result was able to detect epilepsy and enhance the diagnosis of epilepsy with an accuracy of 98.4%. The research was also compared with Artificial Neural Network using Feed-Forward network, the result shows that GOA_ANN approach performed better.

Epilepsy Detection Using Artificial Neural Network and Grasshopper Optimization Algorithm (GOA)

Cloud is known as a highly-available platform that has become most popular among businesses for all information technology needs. Being a widely used platform, it’s also a hot target for cyber-attacks. Distributed Denial of Services (DDoS) is a great threat to a cloud in which cloud bandwidth, resources, and applications are attacked to cause service unavailability. In a DDoS attack, multiple botnets attack victim using spoofed IPs with a huge number of requests to a server. Since its discovery in 1980, numerous methods have been proposed for detection and prevention of network anomalies. This study provides a background of DDoS attack detection methods in past decade and a survey of some of the latest proposed strategies to detect DDoS attacks in the cloud, the methods are further compared for their detection accuracy.

/pdf/ddos-attack-detection-strategies-in-cloud-a-comparative-stud-2371v8hjgh.pdf

DDOS Attack Detection Strategies in Cloud A Comparative Stud

Cloud is known as highly-available platform that has become most popular among businesses for all IT needs. Being widely used platform, it’s also a hot target to cyber-attacks. Distributed Denial of Services (DDoS) is a great threat to cloud in which cloud bandwidth, resources and applications are attacked to cause service unavailability. In a DDoS attack, multiple botnets attack victims using spoofed IPs with a huge number of requests to a server. Since its discovery in 1980, numerous methods have been proposed for the detection and prevention of network anomalies. This study provides a background of DDoS attack detection methods in the past decade and a survey of some of the latest proposed strategies to detect DDoS attacks in the cloud, the methods are further compared for their detection accuracy.

/pdf/a-survey-of-ddos-attack-detection-strategies-in-cloud-4lx7ou3jtx.pdf

A Survey of DDoS Attack Detection Strategies in Cloud

Epilepsy is one of the most common neurological disorders, and afflicts approximately 70 million people globally. 30-40% of patients have refractory epilepsy, where seizures cannot be controlled by anti-epileptic medication, and surgery is neither appropriate, nor available. The unpredictable nature of epileptic seizures is the primary cause of mortality among patients, and leads to significant psychosocial disability. If seizures could be predicted in advance, automatic seizure warning systems could transform the lives of millions of people. This study presents a performance comparison of artificial neural network and support vector machine classifiers, using EEG spectral features to predict the onset of epileptic seizures. In addition, the study also examines the influence of EEG window size, feature selection, and data sampling on classification performance. A total of 216 generalised models were trained and tested on a public seizure database, which contained over 1300 hours of EEG data from 7 subjects. The results showed that ANN outperform SVM, when using spectral features (p = 0.035). The beta and gamma frequency bands were shown to be the best predictors of seizure onset. No significant differences in performance were determined for the different window sizes, or for the feature selection methods. The data sampling method significantly influenced the performance (p < 0.001), and highlighted the importance of treating class imbalance in EEG datasets.

/pdf/a-performance-comparison-of-neural-network-and-svm-13wrotgj4j.pdf

A Performance Comparison of Neural Network and SVM Classifiers Using EEG Spectral Features to Predict Epileptic Seizures

Electroencephalography (EEG) is an important tool in analyzing brain activity. EEG recording is effectively used for detection and prediction of electrophysiological abnormalities due to epilepsy. Epileptic seizure is a brain disorder which affects the patients acutely. Seizures are controllable with medication in 70% of the cases, however the rest may continue to have recurring epileptic seizures despite medications. Since seizures are unpredictable clinically, these patients will also live with perpetual anxiety about the onset of seizure, apart from being affected by the seizure consequences such as drowsiness, headache, vomiting, etc. The seizures can cause injury to the patients, and in some cases may even result in death. Seizure prediction can aid patients with disabling seizure by detecting the seizure precursors in advance and alerting the patients or their caregivers. If the seizure is predicted in advance it can be aborted by fast acting Anti-epileptic drugs (AEDs) or other treatment procedures. This will also aid pre-surgical video EEG monitoring wherein prediction of the ictal onset zone is paramount and machine alarms can be devised. In this paper, we are comparing the results of our research work related to the seizure prediction models. First model, as in usual practice, differentiates between preictal and interictal data segments only. The other seizure prediction model uses Interictal Epileptiform Discharges (IEDs), Electrographic Seizures (ES) and ictal data segments in addition to the first model. We found that the latter one provided better results and improved the seizure prediction performance.

K Parvathy

Papers

Use of electrographic seizures and interictal epileptiform discharges for improving performance in seizure prediction