Pediatric Seizure Prediction in Scalp EEG Using a Multi-Scale Neural Network With Dilated Convolutions

Abstract

Objective: Epileptic seizure prediction based on scalp electroencephalogram (EEG) is of great significance for improving the quality of life of patients with epilepsy. In recent years, a number of studies based on deep learning methods have been proposed to address this issue and achieve excellent performance. However, most studies on epileptic seizure prediction by EEG fail to take full advantage of temporal-spatial multi-scale features of EEG signals, while EEG signals carry information in multiple temporal and spatial scales. To this end, in this study, we proposed an end-to-end framework by using a temporal-spatial multi-scale convolutional neural network with dilated convolutions for patient-specific seizure prediction. Methods: Specifically, the model divides the EEG processing pipeline into two stages: the temporal multi-scale stage and the spatial multi-scale stage. In each stage, we firstly extract the multi-scale features along the corresponding dimension. A dilated convolution block is then conducted on these features to expand our model’s receptive fields further and systematically aggregate global information. Furthermore, we adopt a feature-weighted fusion strategy based on an attention mechanism to achieve better feature fusion and eliminate redundancy in the dilated convolution block. Results: The proposed model obtains an average sensitivity of 93.3%, an average false prediction rate of 0.007 per hour, and an average proportion of time-in-warning of 6.3% testing in 16 patients from the CHB-MIT dataset with the leave-one-out method. Conclusion: Our model achieves superior performance in comparison to state-of-the-art methods, providing a promising solution for EEG-based seizure prediction.

Multi-Channel Vision Transformer for Epileptic Seizure Prediction

“…We also compare our MViT approach with other concurrent and recently-developed deep learning methods that either use CNN [ 23 , 24 , 43 , 47 , 49 , 54 ] or long short-term memory (LSTM) [ 42 , 44 ] for epileptic seizure prediction. In [ 23 ], the raw EEG signals were converted to 3D wavelet tensors (time × scales × channels) and fed into a CNN model, which achieved a prediction sensitivity of 86.6% and a FPR of 0.147/h.…”

“…The results showed that both seizure-prediction sensitivity and FPR were considerably improved to 92.0% and 0.12, respectively. More recently, several studies have demonstrated that CNN-based models can be also effectively applied to raw EEG signals and achieve comparable prediction performance with a sensitivity of 92.0–98.8% [ 47 , 49 , 54 ].…”

“…Our MViT algorithm, on the contrary, extracts the distinguishable EEG features in an automated manner, achieving higher seizure-prediction sensitivity of 99.8%, as shown in Table 1. We also compare our MViT approach with other concurrent and recently-developed deep learning methods that either use CNN [23,24,43,47,49,54] or long short-term memory (LSTM) [42,44] for epileptic seizure prediction. In [23], the raw EEG signals were converted to 3D wavelet tensors (time × scales × channels) and fed into a CNN model, which achieved a prediction sensitivity of 86.6% and a FPR of 0.147/h.…”

Design of an improved soft computing model for EEG classification using MFCC and quad classifier

“…For this assessment, all out 5000 unique sections were separated from this dataset, and were partitioned into 60:40 proportion for preparing and testing individually. Results were assessed regarding exactness, accuracy, review and deferral, and were contrasted and TTFC [4], NNM [5], and LBP TH [9] for approval purposes. The outcomes for exactness can be seen from table 1 as follows,…”

“…A wide collection of EEG portrayal models is proposed by researchers all through the long haul, and all of them vary similar to congruity, exactness, survey, accuracy and concede execution. For instance, work in [2,3,4] analyzes plan of reduced direction set (RISC)-V convolutional Neural Network (CNN) Coprocessor, blend of direct discriminant examination (LDA), k Nearest Neighbor (KNN), support vector machine (SVM), and counterfeit brain organization (ANN) with ordinary spatial model (CSP), and Transfer TSK Fuzzy Classifier (TTFC) for achieving better portrayal results. These models have incredible precision, yet need terms of exactness execution due to their application-unequivocal portrayal characteristics.…”