Jul 2026· International Conference on Control, Decision and Information Technologies· pp. 1595-1600· 0 citations· 17 references
Abstract
This paper presents a comparative study for epilepsy monitoring using EEG signals along two main axes. The first axis consists of comparing the performance of the differentiation technique, which is known to be very important for the study of non-stationarity, with wavelet transform, which is widely used for detecting different brain rhythms. The second part aims to compare the performance of different machine learning algorithms, including k-Nearest Neighbors (k-NN), Decision Trees, Random Forest, and Support Vector Machines (SVM). Used features are statistical and higher order statistics characteristics such as mean, standard deviation, median, Min-Max, Kurtosis and Skewness. We tested our approach on publicly available and widely used datasets in the literature, namely the University of Bonn dataset and the Bern-Barcelona dataset. The experimental results demonstrate the effectiveness of the differentiation method as an important tool for EEG preprocessing, leading to very high performance.
Epilepsy is neurological disorder which is a result of abnormal brain activity which causes repetitive seizures. The analysis of electroencephalogram (EEG) is vital in determining the pattern of epileptic and facilitating clinical diagnosis. Manual interpretation of EEG signals is very cumbersome and time consuming however, because of the high-dimensional and time varying nature of the brain signals. An automated epilepsy detection (EEG) framework is trained based on deep learning. It is a signal preprocessing, time frequency transformation, hierarchical feature extraction based on a hybrid neural architecture that encodes spatial and temporal EEG features. This is followed by the classification of the learned representations with the aim of obtaining the epileptic and non-epileptic brain activity patterns. The experimental assessment shows a better performance than the traditional procedures. The currently proposed model is significantly more accurate (96.7%), sensitive (96.0%), and specific (96.2%), as compared to classical machine learning models like support vector machines (91.2% accuracy) and random forest models (92.6% accuracy). Such results suggest that the improvement in performance of these techniques will be about 4-5 percent compared to the use of conventional techniques. The framework has a high potential of aiding in stable and automatic diagnosis of epilepsy in clinical settings.
Ritu Nagila, Kalaiyarasan R., M. S. et al.· 2026 International Conferenc...· 0 citations
The classification of harmful brain activities (HBA) is crucial in treating and curing many neurological illnesses. Electroencephalography (EEG) signals provide a non-invasive approach for capturing cerebral activity. Harvard Medical School supplied the biggest dataset to date for the HBAC using EEG signals, i.e., the HMS dataset. In this study, we used the HMS dataset. In the proposed model, we transformed raw EEG data into spectrograms. We cleaned, denoised, and eliminated artifacts employing Butterworth and Filtfilt filters. Spectrograms are obtained, and the Discrete Cosine Transform is produced in the subsequent step for Mel-Frequency Cepstral Coefficients (MFCC) computation. We partitioned signals into brief overlapping segments before using DCT. Each frame is transformed from the time domain to the time-frequency domain with the Hamming window method to mitigate edge effects. Filters based on the Mel scale are used. We computed the logarithm of the Mel spectrum. The DCT decreases dimensionality while preserving essential coefficients. The first MFCC derivative, Delta 1, was computed to improve model performance. It shows the change in the original features throughout time. Delta 1 computes the variance between MFCC frames. Delta 2 features are derived in a comparable manner; however, delta_1 is produced from MFCC, whereas delta_2 is derived from delta_1. We used machine learning classifiers with delta features to enhance the model's speed. We proposed an ML-based ensemble for the HBAC task and achieved an accuracy of 94.5%, which is comparable to the state-of-the-art methods in the literature.
D. Shahzad, Irfanud Din, Amina Umer et al.· 2026 6th International Confe...· 0 citations
Epilepsy and migraine are prone to clinical misdiagnosis due to overlapping clinical manifestations, while the limited availability of electroencephalogram (EEG) data further complicates accurate differential diagnosis. To address this challenge, this study proposes a three-class classification framework for epilepsy, migraine, and healthy controls based on EEG signals collected from 36 participants. After preprocessing the raw EEG data with a 0.5–60 Hz bandpass filter, 15 features were extracted, including 12 statistical features and three nonlinear dynamical features. Four machine learning models, namely support vector machine (SVM), random forest (RF), LightGBM, and XGBoost, were systematically evaluated for classification performance. Among them, XGBoost achieved the best overall results, with a test accuracy of 0.90 and superior performance across all major evaluation metrics. Feature importance analysis based on the gain metric of XGBoost identified the Hurst exponent as the most influential feature. Notably, its inter-group distribution differences were highly consistent with the pathological characteristics of epilepsy and migraine. These findings suggest that EEG-based feature fusion combined with machine learning provides an effective strategy for multi-class neurological disease classification, and that the Hurst exponent may serve as a promising biomarker for the differential diagnosis of epilepsy and migraine.
Shiqi Li, Yao Miao· 2026 IEEE International Conf...· 0 citations
The epileptic seizure (ES) is one of the most prominent neurological conditions, whose detection and classification from the electroencephalogram (EEG) signals is crucial for effective diagnosis of seizures, thereby eliminating the detrimental effects associated with it. However, the development of an automated ES detection system is hindered by the non-stationary, non-linear, and high-dimensional nature of the EEG signals, compounded by noise contamination and inter-subject variability. To address these challenges, this paper proposes an automated ES detection framework based on the De-mixing Multivariate Variational Mode Decomposition (D-MVMD) integrated with the Bayesian Optimized Support Vector Machine (BO-SVM). The D-MVMD decomposes multichannel EEG signals into band-limited intrinsic mode functions (BIMFs) while alleviating the correlation between corresponding modes through an ensemble correlation coefficient, while preserving the seizure characteristics from contamination. Multi-domain features capturing temporal, spectral, and non-linear dynamics of seizure activity are then extracted from the de-mixed BIMFs. ReliefF-ranked random forest-based feature selection is employed to find discriminative features, which are subsequently classified using the optimally tuned BO-SVM classifier. Experimental evaluation on the CHB-MIT demonstrates superior performance, with an accuracy of 98.52%, precision of 98.67%, sensitivity of 98.54%, specificity of 98.54%, and F1 score of 0.98. The model is also evaluated on the Siena dataset to assess its robustness across recording sessions of the same patient. Further, it is analyzed with other state-of-the-art methods, confirming its superior mode separation and enhanced seizure detection. Hence, this developed model proves itself to be an effective and robust model for detecting seizures using the multichannel EEG analysis.
T. V. Manju, M. Hota· IEEE journal of biomedical a...· 0 citations
Neurological illness detection using electroencephalograms (EEGs) is not an easy task due to the nature of brain signals, which are not stationary and noisy. In order to obtain the correct multi-class classification, the paper develops a hybrid deep learning framework that includes pre-processing, graph-based feature generation and efficient time modelling. The initial step of cleaning the data is to eliminate any missing data, or artefacts in EEG signals to obtain higher resulting output. A Graph Convoluted Network (GCN) is then used to approximate the functional connectivity between the EEG electrodes and each of the channels is represented as a node in a modelled graph. The GCN models spatial relationships and gets trained on features that are of relevance to the neurological patterns. The Bidirectional Long Short-Term Memory (BiLSTM) network receives features to acquire forward and backward statistical relationships. To enhance the convergence and speed of the weight selection process, the Enhanced Marine Predators Algorithm (EMPA) is applied to avoid falling into the local minima. Proposed model is more effective with higher accuracy of precision ,recall value and F1-score value. The experimental results demonstrating a sufficient recall of spatio-temporal behaviour and classification performance make the proposed model useful to be used in clinical decision support systems.
Dency Flora G, D. R, M. S. et al.· 2026 4th International Confe...· 0 citations
Epilepsy is a long-term neurological disorder marked by recurring seizures resulting from irregular neuronal activity in the brain. Prompt and precise identification of epileptic events from electroencephalogram (EEG) signals is essential for successful clinical diagnosis. This study introduces a Deep Hybrid Neural Network framework that combines Convolutional Neural Networks (CNN) with the Aquila Optimizer (AO) for the automatic detection of epileptic seizures utilizing EEG data in MATLAB. The suggested system initially converts EEG signals into time–frequency spectrograms through Short-Time Fourier Transform (STFT), allowing the CNN to capture advanced spatial–spectral characteristics. The AO algorithm further improves these features by tuning hyperparameters and choosing the most distinguished fea-ture subsets, thereby boosting classification accuracy and decreasing computational overhead. The Bonn University EEG dataset was used to evaluate the model through a 5-fold cross-validation method, attaining an average accuracy of 95.62%, where per-class sensitivity and specificity surpassed 97%. Comparative evaluation showed that the CNN–AO hybrid sur-passed traditional classifiers in terms of accuracy and convergence reliability. These findings demonstrate the effectiveness of the proposed hybrid framework for automated epileptic seizure detection and suggest its potential suitability for future real-time and wearable healthcare applications following further deployment-oriented validation.
Swati Chowdhuri, Tiyasha Mondal· International Journal of Eng...· 0 citations