Multifuse multilayer multikernel RVFLN+ of process modes decomposition and approximate entropy data from iEEG/sEEG signals for epileptic seizure recognition

A predictive model combining connectomics and entropy biomarkers to discriminate long-term vagus nerve stimulation efficacy for pediatric patients with drug-resistant epilepsy

AIMS

To predict the vagus nerve stimulation (VNS) efficacy for pediatric drug-resistant epilepsy (DRE) patients, we aim to identify preimplantation biomarkers through clinical features and electroencephalogram (EEG) signals and thus establish a predictive model from a multi-modal feature set with high prediction accuracy.

METHODS

Sixty-five pediatric DRE patients implanted with VNS were included and followed up. We explored the topological network and entropy features of preimplantation EEG signals to identify the biomarkers for VNS efficacy. A Support Vector Machine (SVM) integrated these biomarkers to distinguish the efficacy groups.

RESULTS

The proportion of VNS responders was 58.5% (38/65) at the last follow-up. In the analysis of parieto-occipital α band activity, higher synchronization level and nodal efficiency were found in responders. The central-frontal θ band activity showed significantly lower entropy in responders. The prediction model reached an accuracy of 81.5%, a precision of 80.1%, and an AUC (area under the receiver operating characteristic curve) of 0.838.

CONCLUSION

Our results revealed that, compared to nonresponders, VNS responders had a more efficient α band brain network, especially in the parieto-occipital region, and less spectral complexity of θ brain activities in the central-frontal region. We established a predictive model integrating both preimplantation clinical and EEG features and exhibited great potential for discriminating the VNS responders. This study contributed to the understanding of the VNS mechanism and improved the performance of the current predictive model.

Partial Discharge Fault Diagnosis in Power Transformers Based on SGMD Approximate Entropy and Optimized BILSTM

Partial discharge (PD) fault diagnosis is of great importance for ensuring the safe and stable operation of power transformers. To address the issues of low accuracy in traditional PD fault diagnostic methods, this paper proposes a novel method for the power transformer PD fault diagnosis. It incorporates the approximate entropy (ApEn) of symplectic geometry mode decomposition (SGMD) into the optimized bidirectional long short-term memory (BILSTM) neural network. This method extracts dominant PD features employing SGMD and ApEn. Meanwhile, it improves the diagnostic accuracy with the optimized BILSTM by introducing the golden jackal optimization (GJO). Simulation studies evaluate the performance of FFT, EMD, VMD, and SGMD. The results show that SGMD-ApEn outperforms other methods in extracting dominant PD features. Experimental results verify the effectiveness and superiority of the proposed method by comparing different traditional methods. The proposed method improves PD fault recognition accuracy and provides a diagnostic rate of 98.6%, with lower noise sensitivity.

RIHANet: A Residual-based Inception with Hybrid-Attention Network for Seizure Detection using EEG signals

Increasing attention is being given to machine learning methods designed to aid clinicians in treatment selection. Therefore, this has aroused a heightened focus on the auto-detect system of epilepsy utilizing electroencephalogram(EEG) data. However, in order for the recognition model to accurately capture a wide range of features related to channel, frequency, and temporal information, it is necessary to have EEG data that is correctly represented. To tackle this challenge, we propose a Residual-based Inception with Hybrid-Attention Network(RIHANet) to achieve automatic seizure detection. Initially, by employing Empirical Mode Decomposition and Short-time Fourier Transform(EMD-STFT) for data processing, it can improve the quality of time-frequency representation of EEG. Additionally, by applying a novel Residual-based Inception to the network architecture, the detection model can learn local and global multiscale spatial-temporal features. Furthermore, the Hybrid Attention model designed is used to obtain relationships between EEG signals from multiple perspectives, including channels, sub-spaces, and global. Using four public datasets, the suggested approach outperforms the results in the most recent scholarly publications.

A ResNet-LSTM hybrid model for predicting epileptic seizures using a pretrained model with supervised contrastive learning

In this paper, we propose a method for predicting epileptic seizures using a pre-trained model utilizing supervised contrastive learning and a hybrid model combining residual networks (ResNet) and long short-term memory (LSTM). The proposed training approach encompasses three key phases: pre-processing, pre-training as a pretext task, and training as a downstream task. In the pre-processing phase, the data is transformed into a spectrogram image using short time Fourier transform (STFT), which extracts both time and frequency information. This step compensates for the inherent complexity and irregularity of electroencephalography (EEG) data, which often hampers effective data analysis. During the pre-training phase, augmented data is generated from the original dataset using techniques such as band-stop filtering and temporal cutout. Subsequently, a ResNet model is pre-trained alongside a supervised contrastive loss model, learning the representation of the spectrogram image. In the training phase, a hybrid model is constructed by combining ResNet, initialized with weight values from the pre-trained model, and LSTM. This hybrid model extracts image features and time information to enhance prediction accuracy. The proposed method's effectiveness is validated using datasets from CHB-MIT and Seoul National University Hospital (SNUH). The method's generalization ability is confirmed through Leave-one-out cross-validation. From the experimental results measuring accuracy, sensitivity, and false positive rate (FPR), CHB-MIT was 91.90%, 89.64%, 0.058 and SNUH was 83.37%, 79.89%, and 0.131. The experimental results demonstrate that the proposed method outperforms the conventional methods.

Automatic seizure detection and classification using super-resolution superlet transform and deep neural network -A preprocessing-less method

CONTEXT

Epilepsy, characterized by recurrent seizures, is a chronic brain disease that affects approximately 50 million. Recurrent seizures characterize it. A seizure, a burst of uncontrolled electrical activity between brain cells, results in temporary changes in behavior, level of consciousness, and involuntary movements. An accurate prediction of seizures can improve the standard of living in epileptic subjects. The increasing capabilities of machine learning and computer-assisted devices can detect seizures accurately with minimal human intervention.

PROPOSED APPROACH

This paper proposes a method to detect seizure and non-seizure events using superlet transform (SLT) and a deep convolution neural network: VGG-19. The electroencephalogram (EEG) dataset from the University of Bonn is used to validate the efficacy of the proposed method.

METHODOLOGY

SLT, a high-resolution time-frequency technique, converts EEG records into two-dimensional (2-D) images. SLT provides a high-resolution time-frequency representation reflecting the oscillation bursts in an EEG record. The time-frequency representations as 2-D images are fed to a pre-trained convolutional neural network: VGG-19. The last layers of VGG-19 are replaced with new layers to accommodate the different classification problems.

RESULTS

The proposed method achieved an accuracy of 100% for all seven seizure and non-seizure detection cases considered in this work. In the case of three and five-class classification problems, the proposed method has better accuracy than other existing methods. The CHB-MIT scalp EEG database is also used to assess the effectiveness of the proposed method, which achieved a classification accuracy of 94.3% in distinguishing between seizure and non-seizure events.

CONCLUSION

The results obtained using the proposed methodology show the efficacy of the proposed method in accurately detecting seizures and other brain activity with the least pre-processing and human involvement. The proposed method can assist medical practitioners by saving their effort and time.

Gradient boosting DD-MLP Net: An ensemble learning model using near-infrared spectroscopy to classify after-stroke dyskinesia degree during exercise

This study aims to develop an automatic assessment of after-stroke dyskinesias degree by combining machine learning and near-infrared spectroscopy (NIRS). Thirty-five subjects were divided into five stages (healthy, patient: Brunnstrom stages 3, 4, 5, 6). NIRS was used to record the muscular hemodynamic responses from bilateral femoris (biceps brachii) muscles during passive and active upper (lower) limbs circular exercise. We used the D-S evidence theory to conduct feature information fusion and established a Gradient Boosting DD-MLP Net model, combining the dendrite network and multilayer perceptron, to realize automatic dyskinesias degree evaluation. Our model classified the upper limb dyskinesias with high accuracy: 98.91% under the passive mode and 98.69% under the active mode, and classified the lower limb dyskinesias with high accuracy: 99.45% and 99.63% under the passive and active modes, respectively. Our model combined with NIRS has great potential in monitoring the after-stroke dyskinesias degree and guiding rehabilitation training.

Epileptic seizure detection with deep EEG features by convolutional neural network and shallow classifiers

Introduction

In the clinical setting, it becomes increasingly important to detect epileptic seizures automatically since it could significantly reduce the burden for the care of patients suffering from intractable epilepsy. Electroencephalography (EEG) signals record the brain's electrical activity and contain rich information about brain dysfunction. As a non-invasive and inexpensive tool for detecting epileptic seizures, visual evaluation of EEG recordings is labor-intensive and subjective and requires significant improvement.

Methods

This study aims to develop a new approach to recognize seizures automatically using EEG recordings. During feature extraction of EEG input from raw data, we construct a new deep neural network (DNN) model. Deep feature maps derived from layers placed hierarchically in a convolution neural network are put into different kinds of shallow classifiers to detect the anomaly. Feature maps are reduced in dimensionality using Principal Component Analysis (PCA).

Results

By analyzing the EEG Epilepsy dataset and the Bonn dataset for epilepsy, we conclude that our proposed method is both effective and robust. These datasets vary significantly in the acquisition of data, the formulation of clinical protocols, and the storage of digital information, making processing and analysis challenging. On both datasets, extensive experiments are performed using a cross-validation by 10 folds strategy to demonstrate approximately 100% accuracy for binary and multi-category classification.

Discussion

In addition to demonstrating that our methodology outperforms other up-to-date approaches, the results of this study also suggest that it can be applied in clinical practice as well.

A 65nm/0.448 mW EEG processor with parallel architecture SVM and lifting wavelet transform for high-performance and low-power epilepsy detection

In recent years, low-power and wearable biomedical testing devices have emerged as a key answer to the challenges associated with epilepsy disorders, which are prone to crises and require prolonged monitoring. The feature vector of the electroencephalographic (EEG) signal was extracted using the lifting wavelet transform algorithm, and the hardware of the lifting wavelet transform module was optimized using the canonic signed digit (CSD) coding method. A low-power EEG feature extraction circuit with a power consumption of 0.42 mW was constructed. This article employs the support vector machine (SVM) technique after feature extraction to categorize and identify epilepsy. A parallel SVM processing unit was constructed to accelerate classification and identification, and then a high-speed, low-power EEG epilepsy detection processor was implemented. The processor design was completed using TSMC 65 nm technology. The chip size is 0.98 mm2, operating voltage is 1 V, operating frequency is 1 MHz, epilepsy detection latency is 0.91 s, power consumption is 0.448 mW, and energy efficiency of a single classification is 2.23 μJ/class. The CHB-MIT database test results show that this processor has a sensitivity of 91.86% and a false detection rate of 0.17/h. Compared to other processors, this processor is more suitable for portable/wearable devices.