Epileptic Seizure Detection Using Multi-Channel EEG Wavelet Power Spectra and 1-D Convolutional Neural Networks

Performance investigation of epilepsy detection from noisy EEG signals using base-2-meta stacking classifier

Epilepsy is a chronic neurological disease, characterized by spontaneous, unprovoked, recurrent seizures that may lead to long-term disability and premature death. Despite significant efforts made to improve epilepsy detection clinically and pre-clinically, the pervasive presence of noise in EEG signals continues to pose substantial challenges to their effective application. In addition, discriminant features for epilepsy detection have not been investigated yet. The objective of this study is to develop a hybrid model for epilepsy detection from noisy and fragmented EEG signals. We hypothesized that a hybrid model could surpass existing single models in epilepsy detection. Our approach involves manual noise rejection and a novel statistical channel selection technique to detect epilepsy even from noisy EEG signals. Our proposed Base-2-Meta stacking classifier achieved notable accuracy (0.98 ± 0.05), precision (0.98 ± 0.07), recall (0.98 ± 0.05), and F1 score (0.98 ± 0.04) even with noisy 5-s segmented EEG signals. Application of our approach to the specific problem like detection of epilepsy from noisy and fragmented EEG data reveals a performance that is not only superior to others, but also is translationally relevant, highlighting its potential application in a clinic setting, where EEG signals are often noisy or scanty. Our proposed metric DF-A (Discriminant feature-accuracy), for the first time, identified the most discriminant feature with models that give A accuracy or above (A = 95 used in this study). This groundbreaking approach allows for detecting discriminant features and can be used as potential electrographic biomarkers in epilepsy detection research. Moreover, our study introduces innovative insights into the understanding of these features, epilepsy detection, and cross-validation, markedly improving epilepsy detection in ways previously unavailable.

Investigating population-specific epilepsy detection from noisy EEG signals using deep-learning models

"Epilepsy is a chronic brain disorder that affects people of all ages. The cause of epilepsy is often unknown and its effect in different age groups is not yet investigated. The main objective of this study is to introduce a novel approach that successfully detects epilepsy even from noisy EEG signals. In addition, this study also investigates population specific epilepsy detection for providing novel insights. Correspondingly, we utilized the TUH EEG corpus database, publicly available challenging multi-channel EEG database containing detailed patient information. We applied a band-pass filter and manual noise rejection to remove noise and artifacts from EEG signals. We then utilized statistical features and correlation to select channels, and applied different transform analysis methods such as continuous wavelet transform, spectrogram, and Wigner-Ville distribution, with and without ensemble averaging, to construct an image dataset. Afterwards, we used various deep-learning models for general analysis. Our findings suggest that different models such as DenseNet201, DenseNet169, DenseNet121, VGG16, VGG19, Xception, InceptionV3, and MobileNetV2 performed better while using images generated from different approaches in general analysis. Furthermore, we split the dataset into two sections according to age for population analysis. All the models that performed well in the general analysis were used for population analysis, which provided novel insights in epilepsy detection from EEG. Our proposed framework for epilepsy detection achieved 100% accuracy, which outperforms other concurrent methods."

Efficient Epileptic Seizure Detection Using CNN-Aided Factor Graphs

We propose a computationally efficient algorithm for seizure detection. Instead of using a purely data-driven approach, we develop a hybrid model-based/data-driven method, combining convolutional neural networks with factor graph inference. On the CHB-MIT dataset, we demonstrate that the proposed method can generalize well in a 6 fold leave-4-patient-out evaluation. Moreover, it is shown that our algorithm can achieve as much as 5% absolute improvement in performance compared to previous data-driven methods. This is achieved while the computational complexity of the proposed technique is a fraction of the complexity of prior work, making it suitable for real-time seizure detection.

EEG Driven Autonomous Injection System For An Epileptic Neuroimaging Application

Seizure episodes are frequently observed for adults and children suffering from medically refractory epilepsy and the events remain debilitating unless treated with a more comprehensive approach. Ictal perfusion studies with single-photon emission computed tomography (SPECT) is one of the non invasive imaging modality that has been extensively used to adequately localize the seizure focus. Current practices include the tracer injection within a short time interval at the onset of seizure to generate desirable SPECT scan quality with accurate information on foci region. However, the onset of a seizure is a highly unpredictable event and also with added subclinical events, the overall procedure makes it difficult to administer the tracer manually within the ideal time frame.Hence a complete autonomous injection of radioactive tracer element without manual intervention is expected to offer a highly accurate epileptical focus region and aids in further management of the patient. Electroencephalogram (EEG) physiological signals in the preictal phase contain sufficient indicators to predict the seizure event. The proposed injection system works on the seizure prediction model from the EEG signals to release the dosage, making the system completely autonomous in action. The accuracy of the prediction model based on the publicly available seizure embedded EEG datasets was designed to achieve 94% accuracy, and the model was deployed on an edge system. The syringe based injection system was characterized to emulate dosage release action with minimum volumetric error, and low injection time, on predicting seizure Ictal event from the EEG signal. The proposed system is a step towards developing an autonomous injection system for epileptic neuroimaging applications in hospital settings.Clinical relevance- Autonomous injection of tracer dosage for obtaining accurate and high quality Ictal SPECT scan results is preferred over the manual operation in clinical and hospital residential settings as a part of pre-surgical evaluations. The EEG signal based early prediction of seizure ensures adequate time for radioactive tracer element to reach the brain cells and eventually helps to accurately localize the onset region of seizure in the brain. The EEG driven automated injection system for the noninvasive Ictal SPECT method is clinically important as a pre-surgical evaluation in MRI negative or discordant cases for further surgical actions.