Detecting abnormal electroencephalograms using deep convolutional networks

Abstract

Objectives: Electroencephalography (EEG) is a central part of the medical evaluation for patients with neurological disorders. Training an algorithm to label the EEG normal vs abnormal seems challenging, because of EEG heterogeneity and dependence of contextual factors, including age and sleep stage. Our objectives were to validate prior work on an independent data set suggesting that deep learning methods can discriminate between normal vs abnormal EEGs, to understand whether age and sleep stage information can improve discrimination, and to understand what factors lead to errors. Methods: We train a deep convolutional neural network on a heterogeneous set of 8,522 routine EEGs from the Massachusetts General Hospital. We explore several strategies for optimizing model performance, including accounting for age and sleep stage. Results: The area under the receiver operating characteristic curve (AUC) on an independent test set (n = 851) is 0.917 marginally improved by including age (AUC=0.924), and both age and sleep stages (AUC= 0.925), though not statistically significant. Conclusions: The model architecture generalizes well to an independent dataset. Adding age and sleep stage to the model does not significantly improve performance. Significance: Insights learned from misclassified examples, and minimal improvement by adding sleep stage and age suggest fruitful directions for further research.

EEG‐based outcome prediction after cardiac arrest with convolutional neural networks: Performance and visualization of discriminative features

“…Finally, our model incorporates EEG data and no other modalities, such as clinical information, somatosensory evoked potentials, biological markers, and MRI findings. Future implementations could include such data, for instance as additional unit in the fully connected layer (van Leeuwen et al, ).…”

“…In particular, convolutional neural networks (CNNs)—a certain type of DL network loosely inspired by the animal visual system, in which connections between (convolutional) layers are made by sliding filters across the input data—have been demonstrated to be extremely efficient for analyzing images (Krizhevsky, Sutskever, & Hinton, ). CNNs and other DL architectures have been applied to EEG data, either for clinical applications such as scoring of sleep stages (Biswal et al, ), detection of focal epileptiform discharges (Johansen et al, ; Tjepkema‐Cloostermans et al, ), detection of “abnormality” or “pathologies” in clinical EEGs (Schirrmeister et al, ; van Leeuwen et al, ), or for brain–computer interfaces (Carvalho et al, ; Lawhern et al, ; Schirrmeister et al, ). Two recent studies used CNNs for prognostication in comatose patients after CA.…”

Radiomics and Machine Learning in Oral Healthcare

“…The reference test was established by two experts in EEG. Regarding the performance, the area under the curve was 0.924 for the developed model …”

“…These imaging features may aid in the early detection of malignant or many bone diseases. They can also be used to predict treatment response to therapies, including oncotherapy, and to estimate functional parameters . The combination of these imaging features with other clinical and genetic data may improve the capacity of detecting and predicting diagnosis and outcomes.…”

Deep Learning in Physiological Signal Data: A Survey

“…Hand motion recognition [9][10][11][12][13][14][15][16][17], Muscle activity recognition [18][19][20][21][22][23] ECG Heartbeat signal classification , Heart disease classification [49][50][51][52][53][54][55][56][57][58][59][60][61][62][63], Sleep-stage classification [64][65][66][67][68], Emotion classification [69], age and gender prediction [70] EEG Brain functionality classification , Brain disease classification , Emotion classification [122][123][124][125][126][127][128][129], Sleep-stage classification [130][131][132][133]…”