Time–Frequency Decomposition of Scalp Electroencephalograms Improves Deep Learning-Based Epilepsy Diagnosis

Abstract

Epilepsy diagnosis based on Interictal Epileptiform Discharges (IEDs) in scalp electroencephalograms (EEGs) is laborious and often subjective. Therefore, it is necessary to build an effective IED detector and an automatic method to classify IED-free versus IED EEGs. In this study, we evaluate features that may provide reliable IED detection and EEG classification. Specifically, we investigate the IED detector based on convolutional neural network (ConvNet) with different input features (temporal, spectral, and wavelet features). We explore different ConvNet architectures and types, including 1D (one-dimensional) ConvNet, 2D (two-dimensional) ConvNet, and noise injection at various layers. We evaluate the EEG classification performance on five independent datasets. The 1D ConvNet with preprocessed full-frequency EEG signal and frequency bands (delta, theta, alpha, beta) with Gaussian additive noise at the output layer achieved the best IED detection results with a false detection rate of 0.23/min at 90% sensitivity. The EEG classification system obtained a mean EEG classification Leave-One-Institution-Out (LOIO) cross-validation (CV) balanced accuracy (BAC) of 78.1% (area under the curve (AUC) of 0.839) and Leave-One-Subject-Out (LOSO) CV BAC of 79.5% (AUC of 0.856). Since the proposed classification system only takes a few seconds to analyze a 30-min routine EEG, it may help in reducing the human effort required for epilepsy diagnosis.

Multi-center Assessment of CNN-Transformer with Belief Matching Loss for Patient-independent Seizure Detection in Scalp and Intracranial EEG

“…For all the EEGs, a 4 th order Butterworth notch őlter at 60Hz (USA) and 50Hz (EU) is applied to remove electrical interference 65 . Next, a 1Hz high-pass őlter (4 th order) is implemented to reject DC shifts and baseline ŕuctuations 66 . Finally, all the EEGs are downsampled to a sampling frequency F s of 128Hz.…”

“…We perform seizure detection őrst at individual channels (channel-level detection), next at multi-channel segments (segment-level detection), and at last, we detect the start and end points of the seizures in the entire multi-channel EEG (EEG-level detection) [65][66][67] (see Supplementary Figure 1). The pipeline of the proposed seizure detector is displayed in Figure 3.…”

Does Deep Learning Have Epileptic Seizures? On the Modeling of the Brain

“…But in any case, the differentiation between normal and pathological in human epileptic seizures cannot be transposed to the question of normal and pathological in DL. DL can only be a seizures classifier, detector, or eventually be used for prediction, given some specific input data [16][17][18].…”

Deep learning methods for diagnosis of epilepsy from electroencephalograms

“…Several works perform EEG artifact rejection, each focusing on different areas of artifacts. Common methods for artifact detection and rejection includes common average reference (CAR) [272], independent component analysis (ICA) [199], high amplitude rejection [218,273,274], wavelet transforms [275], CNN [223], and LSTM [276]. We summarized existing literature on artifact detection and rejection in Table 2.2.…”

“…For all the EEGs, a 4 th order Butterworth notch filter at 60Hz (USA) and 50Hz (EU) is applied to remove electrical interference [218]. Next, a 1Hz high-pass filter (4 th order) is implemented to reject DC shifts and baseline fluctuations [274]. Finally, all the EEGs are downsampled to a sampling frequency F s of 128Hz.…”